My 2024 Artificial Intelligence (AI) Quest
With forty years in computer technology, I’ve witnessed a
series of paradigm shifts, a fundamental change in the basic concepts and
practices of data processing and information management. Some were more
dramatic that others;
Punched card, data-storage-media processing, and the
migration from unit-record to programmable computers, was the predominate
computer technology of the sixties.
Soon, came magnetic data storage media, on computers, like
tape and magnetic/optical disk.
Then came, data on-line communication and wired, and
wireless networks.
I never though Personal Computers would usher-in that huge
information management PC-network and paradigm shift.
Broadband Internet, Cloud Computing, and that family of
storage exploded the market.
Next, the Mobile shift put it all in our pocket phones.
Jump to now, and Artificial Intelligence has EXPLODED on
the scene. It started in academic and corporate research labs and blasted into
the commercial market.
So, over forty years, I’ve learned that the publicity
outpaces the reality of a technology, because adaptation takes time, BUT AI is
moving much faster than expected. Computer professionals would sometimes delay
implementing new technologies, letting others suffer the bugs, but with AI,
this is not an option for users. This paradigm shift in AI computer technology
is the most dramatic in my experience. Every aspect of AI is new and very
complex, with concepts, functions, features, terminology, hardware, software,
and firmware, all a new-fangled environment. Initially, my AI research was very
frustrating; the more I studied, the more confusing it became. Over time, I began
to make progress and now it’s time for some formal education.
The seven LinkedIn-example stages of AI
1. Rule-Based AI Systems
Description: In this stage, AI systems follow pre-set rules
defined by human programmers. These systems are limited to specific tasks
without the ability to adapt or learn.
Example: Early expert systems like MYCIN, developed in the
1970s to diagnose bacterial infections, followed a strict set of medical rules.
Another Example: TurboTax is a modern example, helping
users file taxes based on rule-based logic but lacking adaptability or learning
capabilities.
2. Context Awareness and Retention Systems
Description: AI at this stage can understand and retain
context from past interactions to influence future decisions, offering a more
personalized user experience.
Example: Amazon Alexa or Google Assistant can remember user
preferences, adjusting responses based on context. For instance, Alexa might
remind you of a shopping list item mentioned earlier or remember your home
location for better suggestions.
Another Example: Tesla Autopilot uses memory from past
driving experiences to improve performance over time, like identifying frequent
routes.
3. Domain-Specific Mastery Systems
Description: These AI systems specialize in a single domain
and perform at expert levels, far surpassing human capabilities in that
specific field.
Example: IBM’s Watson mastered trivia to win Jeopardy!,
using natural language processing to understand and answer questions.
Another Example: DeepMind’s AlphaGo became a
domain-specific master by defeating world champion Go players, learning complex
strategies via deep learning.
4. Thinking and Reasoning AI Systems
Description: AI systems here can simulate human thought
processes, reasoning through complex tasks, solving novel problems, and making
decisions without direct programming.
5. Artificial General Intelligence (AGI)
Description: AGI would be capable of performing any
intellectual task that a human can. It would understand, learn, and apply
knowledge across a wide array of tasks.
Example: Although AGI is still theoretical, platforms like
DeepMind’s Gato aim to move closer to this stage, as it can perform over 600
different tasks, ranging from robotics to image recognition and text
generation.
6. Artificial Superintelligence (ASI)
Description: ASI represents a system with cognitive
abilities far beyond human intelligence. It could solve global-scale problems,
from climate change to disease, with unimaginable speed and precision.
Example: This stage is still speculative, but imagine AI
models like DeepMind’s AI for scientific discovery, potentially solving complex
problems in physics or medicine.
Another Example: The concept of ASI often relates to future
AI like GPT-5 or beyond, where machines could autonomously innovate in ways
humans may not be able to foresee.
7. AI Singularity
Description: The AI Singularity refers to a hypothetical
future point where AI growth becomes uncontrollable and irreversible,
fundamentally transforming civilization. This point would likely be driven by
an ASI that could improve its own intelligence beyond human comprehension.
Example: This stage is highly speculative, but Ray
Kurzweil’s predictions for the 2040s involve AI surpassing human intelligence,
leading to profound changes in society, economics, and even human biology. This
sounds like; “why would you do this?” but basic economics say; “humans always
prefer more of a good over of a less good”, so the good will outweigh the bad!
The path of AI from rule-based systems to superintelligence
showcases the field’s dynamic nature, full of both immense promise and complex
ethical considerations.
The seven stages of AI are all experiencing tremendous
research, development, and advancement. For sure, the first three stages are in
extensive use today. Generative AI and ChatGPT are producing text, imagery,
audio, video, and data for us.
Large Language Models use massive data to generate accurate
and real responses to our verbal prompts.
The BIG stage 7 is providing a method where us humans can
connect our minds, sort of like computers on the Internet, and instantly, we be
(selectively)sharing our thinking, without our five senses, on the
People-Network. Neuralink, the brain-computer interface and neuroprosthetics
company, started by Elon Musk and associates, is developing ultra-high
bandwidth brain-machine interfaces to connect humans and computers.
My main interest, at this time, is to understand and master
the AI benefits/risks to senior citizens. Currently, we seniors are affected by
the technology but really have no voice in the technology. No doubt, AI is the
biggest computer technology paradigm shift, ever, so far. Experts say;
“The development of full artificial intelligence could
spell the end of the human race.” – Stephen Hawking
“AI is likely to either be the best or worst thing to
happen to humanity.” – Elon Musk
“I am in the camp that is concerned about
superintelligence. But I don’t think we need to be fatalistic about it.” – Bill
Gates
“AI will be part of every industry, enhancing our abilities
in ways we can’t even imagine yet.” – Jeff Bezos
“All AI things considered, Critical Thinking is now a
prerequisite for EVERYONE, because of TRUE/FALSE STUFF in Social Networks and
Artificial Intelligence.” -- ME SAY
Critical thinking is the analysis of available facts,
evidence, observations, and arguments in order to form a judgement, by the
application of rational, skeptical, and unbiased analyses and evaluation.
Nothing new, but not real extant in people!
So, AI is now business-driven by profit-motivation and we
don’t know or didn’t vote for, and we have no control over it. At least, we
should seek out the best information, using our steaming services,
documentaries, publications and any reliable information that we can find, so that
we know what the experts are arguing about. We need AI that AI Should Augment
Human Intelligence, NOT Replace It!
The majority of the current AI users that I talk to, like
it a lot. My personal experience is also very favorable and my main
concentration is representing senior citizens in how AI will affect us.
After a summer-long, world-wide search/research quest, my best
Introduction Course is found at Oxford University, UK. I’m taking this course,
online, this winter. The Oxford Department for Continuing Education has
provided the syllabus, or outline for the course. In advance preparation, I’ve
expanded the bulleted topics with my personal research, in an attempt to get a
leg-up on what the course will teach. My search/research is based on the most
reliable and expert-based information and my notes are extracted from this
information. Sharing this expanded-outline, is my attempt in sharing what
fundamentally, we all need to know and consider, in simple, unassuming
language, as we enter the AI world. Yes, even seniors need to, at least, read
the overview, because you’re going to be a USER!
artificial intelligence, n.
The capacity of computers or other machines to exhibit or
simulate intelligent behavior; the field of study concerned with this. Source:
Oxford English Dictionary
Artificial Intelligence (AI) has become ingrained in the
fabric of our society, often in seamless and pervasive ways that may escape our
attention day-to-day. The ability of machines to sense, process information,
make decisions and learn from experience is a transformative tool for
organizations, from governments to big business. However, these technologies
pose challenges including social and ethical dilemmas.
· This
course provides an essential introduction to the key topics underpinning AI,
including its historical development, theoretical foundations, basic
architecture, modern applications, and ethical implications. The course
investigates the future trajectory of AI and considers its potential for
improving the world while highlighting pitfalls and limitations. It is aimed at
a general audience, including professionals whose work brings them into contact
with AI, and those with no prior knowledge of AI. The course aims to confer an
appreciation of the ways in which our world has already been transformed by AI,
to explain the fundamental concepts and workings of AI, and to equip us with a
better understanding of how AI will shape our society, so we can converse
fluently in the language of the future. In preparation for the course, and Guided
by the syllabus/outline, I’ve added my research findings (+) under the bullets
(.) of the outline (below).
University of Oxford
Department for Continuing Education
Introduction to Artificial Intelligence (Syllabus)
https://conted.ox.ac.uk/courses/introduction-to-artificial-intelligence-online
onlinecourses@conted.ox.ac.uk
Unit 1: What is intelligence?
- The concept of intelligence
+The
ability to acquire and apply knowledge and skills. Our family units have always
taught and guided our offspring to learn and improve on intelligence, making
our world better.
- What is artificial intelligence?
+ Artificial intelligence (AI) is the ability
of a computer or robot to perform tasks commonly associated with intelligent
beings. Most of us don’t realize how much we have already accepted, adopted and
use these tools. We need to realize that the rapid advancements in artificial
intelligence (AI) and machine learning (ML) technologies have led to
significant societal implications. These technological innovations (both
promising and uncertain) have the potential (controlling and dangerous) to
revolutionize various aspects of society, including the future of learning,
social impact, and the nature of work.
- Weak vs Strong AI
+ Strong
or Artificial General Intelligence (AGI) AI refers to a hypothetical machine
that exhibits human cognitive abilities. It can tackle diverse problems and
develop new approaches to solve the task. Strong AI aims to create machines
with human-like cognitive abilities, self-awareness, and adaptability.
+ Weak
or Narrow AI (being Rules-Based) refers to the use of advanced algorithms to
accomplish specific problem solving or reasoning tasks that do not encompass
the full range of human cognitive abilities. Rule-based AI systems operate on a
set of predefined rules created by human experts. These rules dictate the
system's behavior in response to specific inputs. (Alexa, Chatbot, Amazon,
Spotify, Your self-driving car). Weak AI (our current model) can outperform
humans on the specific tasks it is designed for, but it operates under far more
constraints than even the most basic human intelligence.
- A Brief History of AI
+The
field of AI wasn't formally founded until 1956, at a conference at Dartmouth
College, in Hanover, New Hampshire, where the term " artificial
intelligence " was coined. As of 2024, AI is surpassing human performance
on numerous benchmarks, including simulate conversation with human users, image
classification, visual reasoning, and English understanding thru text or voice.
My personal experience with Genie Chatbot polishes my writing, and vastly
improves my communication and productivity skills. We all use other Chatbots,
for searches, translation, shopping, and a wide variety of helpful tasks.
- The Golden Age of AI
+The
term “golden age of AI” is often used to describe the current era of rapid
advancements and widespread adoption of artificial intelligence technologies
(Major computer technology paradigm shift). This period is marked by
significant breakthroughs in areas like generative AI, machine learning, and
natural language processing. Tools like ChatGPT have captured the public’s
imagination and are transforming personal and corporate computing. One of the
most compelling reasons to study AI is to learn the ethical implications that
come with advancing technologies.
·
Applications of AI
+Artificial
Intelligence (AI) has a wide range of applications across various business
sectors. (Medical, Education, Government, Commercial, Manufacturing,
Industrial, Military, Aerospace, and …..)
Unit 2: Artificial intelligence and society
- Data governance
+Data
governance is critical in quality assurance for ethical AI by establishing
frameworks and guidelines that govern the collection, storage, usage and
sharing of data. It ensures that the data used to train AI models is accurate,
reliable and helps mitigate bias. (Like other (new) computer technologies,
self-governing will be an issue.)
- AI and Equality
+AI
has the potential to increase inequality. As AI is increasingly applied to make
consequential decisions that affect social, political, and economic rights, it
is imperative that we ensure these systems are built and applied in ways that
uphold principles of fairness, accountability, and transparency. (This is why
Seniors need a voice, like other business sectors). Also, young users can be
led like sheep, because commonly, their minds are not developed to discriminate
about information. Plus, many adult users do not possess critical thinking
skills and are lead by deceptive media.
- AI and employment
+Overall,
while AI poses challenges, it also offers opportunities for innovation and
growth in the job market. For example: We have a manufacturing plant of 1600
employees that AI has reduced to 400 employees. Now, the staff is increasing
with new tech jobs. AI will create more jobs, so the candidates must (as usual)
qualify themselves, with the necessary skills.
- Economic opportunities of AI
+AI
is already affecting how economies grow, produce jobs, and trade
internationally. The myriad ways
globalization impacts our lives are connected to AI and we tend to blame our
President and government for economic pain, when in fact, the answer is very
complex. Critical thinking is a new requirement for us all to learn, UNDERSTAND,
and adopt the inevitable changes and new technologies.
- Risks of AI
+AI
has the potential to revolutionize various fields, but also poses serious
threats to society and humanity. (Lots known and unknown here). Some people
feel that they will fall behind as AI becomes more prevalent. So, is AI
something we should be scared of? The fears of AI seem to stem from a few
common causes: general anxiety about machine intelligence, the fear of mass
unemployment, concerns about super-intelligence, putting the power of AI into
the wrong people’s hands, and general concern and caution when it comes to new
technology. Artificial intelligence algorithms will soon reach a point of rapid
self-improvement that threatens our ability to control them and poses great
potential risk to humanity, and many AI experts are very concerned that it
could be catastrophic. We’re using it and liking it but it will usher in dramatic
change, is it going (progressing) too fast? Do you think that AI could control
us, like we’re just robots. Certainly, our cooperate oligopolies are going to
use AI to increase profit and simultaneously raise prices even more. Then,
there’s the China vs USA situation, where China is winning the computer
technology race. China is intent on dominating and monopolizing Quantum
Computing and Artificial Intelligence. America has created the Special
Competitive Studies Project to address and strengthen America’s long-term
competitiveness. I can’t even imagine how all this will play out. Will we be
able to validate and check our personal data that AI will capture/generate. In
a recent meeting with young computer technicians, I just discovered that
they’re concerned about the computer tech jobs that AI is displacing. This
whole technology this is huge and vey complex, growing and changing every day,
so we must be aware of how it will affect (physiological economic, political)
us senior citizens.
+ Automation-spurred
job loss (A fact, and coming at the worst time in a divided nation). Numerous
observers believe recent developments in robotics and AI may cause an
unprecedented wave of automation-related job losses. The answer is get prepared
for AI and develop those AI skills, because it’s a new, unavoidable field.
+ Deepfakes
Deepfakes
use AI to replace the likeness of one person with another in image, video or
audio. Now, AI can make anyone into any image, and vice versa, and celebrities
(or anyone) have lost control of their likeness/image on social media etc. Don’t take anything at “FACE-VALUE”, you MUST
know how to spot deepfakes! You must learn critical-thinking skills and not
believe anything (at face value) that you see!
+ Privacy
violations
Artificial
intelligence has been no different when seen through a privacy-by-design lens,
as privacy has not been top-of-mind in the development of AI technologies.
There is a high risk to individuals’ rights and freedoms in the AI processing
of personal data, something quite different to the risk posed by data breaches,
but also with very little “fallout” for the companies responsible. Some privacy
challenges of AI include:
Data
persistence – data existing longer than the human subjects that created it,
driven by low data storage costs
Data
repurposing – data being used beyond their originally imagined purpose
Data
spillovers – data collected on people who are not the target of data collection
Just
recently, the largest data spillover in history occurred, where billions of people
(targets or non-targets of data collection) were data-scraped by National
Public Data, a company most have never heard of and have done no business with.
This background-check company, grabs data from every source they can and sells
it, without anyone’s permission or authorization, and claims no responsibility
for the security of the scraped-data (to me it’s stolen data). Its all about
money;3 same is true for networks, no guarantees!
+ Algorithmic
bias caused by bad data
Algorithmic
bias is when processes commit systematic errors that unfairly favor or
discriminate against certain groups of people. These AI biases are the result
of poor training data and the biases of the humans who compiled the data and
trained the algorithms. We’ll be trusting humans that have human flaws, to do
this, so we better be able to critically think, as users.
+ Socioeconomic
inequality
Most
empirical studies find that AI technology will not reduce overall employment.
However, it is likely to reduce the relative amount of income going to
low-skilled labor, which will increase inequality across society. Nothing new
for technology, because change is inevitable and our skills must change. Now,
personally, AI is progressing at the worst time ever, because social media has
given us all a voice, and the social divide will intensify. Ideally, we need
good employment opportunities for everyone, depending on their education and
skills. This is a big order!
+Market
volatility
AI is
being increasingly used to analyze and predict market volatility. Techniques
such as artificial neural networks and machine learning algorithms have shown
promise in accurately forecasting stock prices and identifying changes in all
market trends. Yes, it’s loaded with good and bad! The good is how AI shows
promise in taming market volatility. The Bad is the regular Bad, plus unknown,
future developments.
+ Weapons
automatization
AI
weapons automation refers to the use of artificial intelligence in military
applications. Here are some key points about this topic:
Current
AI-enabled weaponry is not yet fully autonomous, but the technology exists.
Advances
in AI empower autonomous weapons and platforms to carry out more sophisticated
behaviors and activities. AI can be used for analyzing the battlefield,
providing augmented reality information to soldiers, and identifying threats.
Deployment
of AI-controlled drones that can make autonomous decisions about killing human
targets is being developed by countries, including the US, China, and Israel.
With
Russia’s invasion of Ukraine as the backdrop, the United Nations recently held
a meeting to discuss the use of autonomous weapons systems, commonly referred
to as killer robots.
- Uncontrollable self-aware AI
+Uncontrollable
self-aware AI is a topic of concern. While it may sound like science fiction,
there are already machines that perform tasks independently without programmers
fully understanding how they learned it. A recent study suggests that it is
virtually impossible to keep a artificial-super-intelligent (ASI) in or under
control. Although there is no theoretical barrier for AI to reach
self-awareness, the conclusion is that we currently cannot control it. (Totally
baffled by this)
·
AI and accountability
+Accountability
in AI refers to the expectation that organizations or individuals will ensure
the proper functioning of the AI systems they design, develop, operate or
deploy, in accordance with their roles and applicable regulatory frameworks.
Providing accountability for trustworthy AI requires that actors leverage
processes, indicators, standards, certification schemes, auditing, and other
mechanisms to follow these steps at each phase of the AI system lifecycle.
There is a growing concern about an "accountability gap" in AI, and
this gap prevails through the entire history of computer technology. Think
about your company; security and accountability costs money, and it’s probably
all about the money, so nah, we don’t need that!
Unit 3: Systems and agents
- Concept of an agent
+An
AI agent is a software entity that uses artificial intelligence techniques to
perceive its environment, make decisions, and take actions to achieve specific
goals. AI agents can operate autonomously or semi-autonomously and are designed
to solve problems, automate tasks, or provide services in various domains.
+ The
Key Characteristics of AI Agents:
+1.
**Perception**: AI agents can gather information from their environment through
sensors or data inputs. This could include anything from visual data from
cameras to numerical data from sensors.
+2.
**Decision Making**: Based on the information they perceive, AI agents use
algorithms and models to analyze data, evaluate options, and make decisions.
This may involve machine learning, rule-based systems, or optimization
techniques.
+3.
**Action**: Once a decision is made, the AI agent takes action to affect its
environment or achieve its objectives. This could involve sending commands to
other systems, generating responses, or interacting with users.
+4.
**Autonomy**: Many AI agents are designed to operate with a degree of
independence, meaning they can perform tasks and make decisions without human
intervention.
+5.
**Learning**: Some AI agents are capable of learning from their experiences,
allowing them to improve their performance over time. This can involve
adjusting their strategies based on feedback or new data.
+The Types
of AI Agents:
+-
**Reactive Agents**: These agents respond to specific stimuli from the
environment but do not have memory or learning capabilities.
+-
**Deliberative Agents**: These agents maintain an internal model of the world
and can plan actions based on that model.
+-
**Learning Agents**: These agents can learn from their experiences and improve
their performance over time, often using machine learning techniques.
+-
**Multi-Agent Systems**: In some applications, multiple AI agents can work
together or compete with each other to achieve individual or collective goals.
- Applications:
+AI
agents are used in various fields, including:
+-
**Customer Support**: Chatbots and virtual assistants that handle customer
inquiries.
+-
**Robotics**: Autonomous robots that can navigate and perform tasks in
real-world environments.
+-
**Gaming**: Non-player characters (NPCs) that interact with players and adapt
to their strategies.
+-
**Recommendation Systems**: Agents that suggest products or content based on
user preferences and behaviors.
+Overall,
AI agents are a fundamental part of the broader landscape of artificial
intelligence, enabling more intelligent, responsive, and autonomous systems
across various applications.
+Structure
of an agent
To
understand the structure of Intelligent Agents, we should be familiar with
Architecture and Agent programs. Architecture is the machinery that the agent
executes on. It is a device with sensors and actuators, for example, a robotic
car, a camera, and a PC. An agent program is an implementation of an agent
function. An agent function is a map from the percept sequence (history of all
that an agent has perceived to date) to an action.
+Rationality
of an agent
To be
considered a rational agent, an AI agent must select actions that maximize its
performance measure for all possible percept sequences. Rationality goes beyond
simply being an agent; it focuses on achieving the desired outcomes given the
available information and prior knowledge.
+Perfect
agents
A
perfect AI agent possesses the following characteristics:
Autonomy:
It can act independently without continual human intervention.
Learning
and Adaptation: It can learn from data and adapt its behavior over time.
Interaction:
It can meaningfully interact with other agents or systems.
Perception:
It has sensors or mechanisms to observe and perceive its environment.
Reasoning
and Decision-Making: It can process information, reason about goals, and make
decisions to achieve those goals.
+Task
environments
In
the context of AI, the task environment refers to the surroundings or
circumstances in which an AI system functions. It includes the physical
environment, digital platforms, and virtualized worlds where AI models and
algorithms are used. The task environment can be classified based on various
factors such as observability, agents, determinism, episodic nature, and
continuity. Back to the computer part, it all about the hardware, firmware,
software, and data.
+Designing
agents
An
artificial intelligence (AI) Designing agent refers to a system or program that
is capable of autonomously performing tasks on behalf of a user or another
system by designing its workflow and utilizing available tools.
+Simple
reflex agent
Simple
reflex agents are fundamental constructs in artificial intelligence (AI) that
perform on a simple precept: they make choices primarily and completely based
on immediate environmental stimuli and then take the appropriate action. This
is like our home thermostat, health-monitoring devices, automobile sensors, and
a plethora of constantly-growing list of other AI stuff. This is the simple
reflex agent we’re all readily adopting, but wait, these agents could do a lot
more! Good or bad?
+Model-based
reflex agents
A
model-based reflex agent is an agent that uses a remembered history of
perceptions to choose actions and form a more comprehensive view of their
environments. Unlike simple reflex agents, model-based reflex agents are
model-based, which means they have knowledge of how things happen in the world.
They take into account both the current percept and an internal state
representing the unobservable aspects of the environment. They update their
internal state based on how the world evolves independently and how their
actions affect the world
+Goal-based
agents
Goal-based
agents are AI systems designed to achieve specific objectives or goals. Unlike
simple reflex agents that act solely based on current perceptions, goal-based
agents consider future consequences of their actions, ensuring that they align
with the set objectives. Given a plan, a goal-based agent attempts to choose
the best strategy to achieve it based on the environment. WE also have readily
adopted this one, in natural language processing. Yes, it’s where you talk to
Alexis, Google, Genie and you-name-it, there’re everywhere.
+Utility-based
agents
The
agents which are developed having their end uses as building blocks are called
utility-based agents. When there are multiple possible alternatives, then to
decide which one is best, utility-based agents are used. They choose actions
based on a preference (utility) for each state. Sometimes achieving the desired
goal is not enough. We may look for a quicker, safer, cheaper trip to reach a
destination. Agent happiness should be taken into consideration. Utility
describes how “happy” the agent is. Because of the uncertainty in the world, a
utility agent chooses the action that maximizes the expected utility. A utility
function maps a state onto a real number which describes the associated degree
of happiness.
Now
folks, by now, you begin to see that AI, like our previous computer technology
paradigm shifts, is way more than anything we’ve ever experienced and it is
incumbent on us to be informed on what’s behind the tools we choose/use. Critical thinking, it’s REQUIRED everywhere!!!!
Unit 4: Logic and language
- Early ideas: Logic and language
+Alan
Turing, at a time when computing power was still largely reliant on human
brains, the British mathematician Alan Turing imagined a machine capable of
advancing far past its original programming. To Turing, a computing machine
would initially be coded to work according to that program and surpass human
capability.
The “Evolution
of Artificial Intelligence” by Joshua Cena, of the University of Manchester,
delves into the rich history and evolutionary journey of artificial
intelligence (AI), tracing its origins from early conceptualizations to its
current applications in various domains. Through a comprehensive review of key
milestones, breakthroughs, and influential figures, the document highlights the
pivotal moments that have shaped the development of AI over time. It explores
how AI has transitioned from theoretical frameworks and symbolic reasoning
approaches to the era of machine learning, deep learning, and neural networks,
leading to transformative advancements in areas such as robotics, natural
language processing, computer vision, healthcare, and other autonomous systems.
- Imitating mathematical intelligence
+AI
has made significant strides in imitating mathematical intelligence,
particularly in solving complex problems that require advanced reasoning skills
and computational power. These advancements are not just about solving math
problems for competitions, this progress indicates that AI is getting closer to
human-like reasoning abilities, which could lead to more powerful AI tools for
scientific research and education
- Propositional logic
+Propositional
logic is a fundamental building block in AI, serving as the language in which
we express knowledge and information in a structured manner. This system allows
us to represent the world's knowledge, facts, and relationships using simple,
atomic propositions, and logical operators like "AND,"
"OR," and "NOT." Propositional logic is based on
propositions, binary statements about the world, that can be either true or
false.
- Designing mathematical languages
+There
are three types of theory language used in designing AI products: formal,
computational, and natural.
1.Formal
languages, such as mathematics, logic, and programming languages, have fixed
meanings and no actual-world semantics.
2.Computational
Languages: These languages refer to real-world entities, events, and thoughts.
They have actual-world references and semantics, making them context-sensitive.
Computational languages are used to model and simulate real-world phenomena within
AI systems.
3.
Unlike formal and computational languages, natural languages (like talking
English, French, Spanish, Japanese, ect.) are dynamic, creative, and
productive. They can refer to an unlimited number of objects and their
attributes across various domains. Natural languages are often used in AI for
tasks involving human-computer interaction, such as natural language processing
(NLP). We’re all currently doing this, much more than we realize.
- Gödel's incompleteness
+Gödel's
incompleteness theorems are two theorems of mathematical logic that are
concerned with the limits of provability, in formal axiomatic theories. Not
sure, but it might be yeas, no, or maybe. This is currently over my head!
- Solving mathematical problems with AI
+
Need more info. This may be about how AI agents can work cooperatively to pass
along problems to another agent, whose environment can best process the meet
the requirements of the query.
- Halting problem
+The
Halting Problem, a fundamental concept in computer science and artificial
intelligence, poses intriguing questions about the limits of computation. It
delves into the feasibility of determining whether a program will eventually
halt or continue to run indefinitely. (Programmers defined this as; hung in a
loop) Logic is limited by the binary bit pattern of the program. Example; an eight-bit
byte is 2^8=256 options A word is a group of bytes. Therefore, a word can be powers
of 2 for 16 bits, 24 bits, 32 bits, and so on.
Unit 5: Expert systems
- What are expert systems?
+An
expert system in artificial intelligence (AI) is a computer system that
emulates the decision-making ability of a human expert. It is designed to solve
complex problems by reasoning through bodies of knowledge, (neural networks)
represented mainly as if–then-else RULES rather than through conventional
procedural code
- Representing knowledge
+Knowledge
representation is the method by which information is formalized for AI systems
to use. It encompasses a variety of techniques designed to represent facts,
concepts, and relationships within a domain, allowing machines to process and
utilize this information effectively.
Primary
goals of knowledge representation;
+Expressiveness:
The ability to represent a wide variety of knowledge.
Efficiency:
The capability to manipulate and reason with knowledge quickly.
Types
of AI knowledge;
Declarative
Knowledge: Facts and information about objects, events, and their
relationships. For example, “Paris is the capital of France.”
Procedural
Knowledge: Knowledge of how to perform tasks. For example, “How to ride a
bicycle.”
Meta-Knowledge:
Knowledge about another knowledge. For example, “The reliability of a source.”
Heuristic
Knowledge: Rules of thumb or best practices. For example, “If the weather is
cloudy, it might rain.”
This
gets way deep, so I’m stopping here, for now!
Understandability:
The ease with which humans can comprehend the represented knowledge.
Scalability:
The ability to handle increasing amounts of knowledge without significant
performance degradation.
- Reasoning with logic
+Reasoning
in AI refers to deriving new information from existing information using
logical rules and principles. AI systems use Reasoning to make inferences, draw
conclusions, and solve problems. Automated reasoning lies at the core of
artificial intelligence, where the focus is on crafting systems that can
independently navigate the realm of logical deductions and inferences. It can
be thought of as giving machines the ability to think logically. Artificial
neural networks are composed of layers of nodes
Each
node is designed to behave similarly to a neuron in the brain
The
first layer of a neural net is called the input layer, followed by hidden
layers, then finally the output layer
Each
node in the neural net performs some sort of calculation, which is passed on to
other nodes deeper in the neural net.
Deeper
we go, we find Artificial Neural Network (ANN) and Biological Neural Network
(BNN), which is another Data Science Course!
- Backward chaining
+Backward
chaining is an inference method where an AI system starts with a goal or
desired outcome and works backward through a series of rules and conditions to
find the necessary steps or conditions to achieve that goal. It’s like solving
a puzzle in reverse, beginning with the solution and tracing back to the
initial conditions. This reminds me of a GIS expert that I worked with. He
said, “play the movie backwards”, to solve/execute your planning, organizing,
management, and development of a data system. Being IMB-trained, I learned
top-down, hierarchy, in business systems. Backward-chaining is, therefore, a
very difficult concept for me!
- Advantages and disadvantages of expert systems
+ This system is heavily dependent on a good
system base. Experts keep updating the information in the knowledge base, and
non-expert makes use of this information for complex problem-solving. OK. Think
about your work experience, some folks are just not very dependable for keeping
a good system base. It really boils down to “can you trust this information”?
So, just like the Internet users, people must have (learn) good
critical-thinking skills.
Unit 6: Connectionist models
- Biological neural network
+A
biological neural network (BNN) is a physical structure found in brains and
complex nervous systems, consisting of interconnected neurons connected by
synapses. These networks allow for communication and information processing
within the nervous system. Neurons are connected by axons and dendrites, and
neurotransmitters are released at synapses to excite or inhibit adjacent
neurons. Scientists
have fused brain-like tissue with electronics to make an ‘organoid neural
network’ that can recognize voices and solve a complex mathematical problem.
Their invention extends neuromorphic computing – the practice of modelling
computers after the human brain – to a new level by directly including brain
tissue in a computer. This seems way out-yonder, but it’s on the AI plan!
- ANNs in action
+An
artificial neuron network (ANN) is a computing system patterned after the
operation of neurons in the human brain. The layered ANN, inspired by the BNN
is the foundation of the AI system, with an Input layer, hidden layers, and an
output layer. Entering the Input layer, a query will be resolved(answered), and
the exit the Output layer. All the AI stuff WE do is facilitated by specialized
algorithms (look up classes of algorithms) which pass through interconnected (one-byte
or 8-bit) nodes of the neural network layers, from the input layer, through the
hidden layers, and finally, to the output layer, with our answer.
- Building blocks of a neural network
+ The
logic gates and electronic circuits of the ANN layers, which support the
artificial intelligence process.
- An example of a neural network (We’re using these
without any real concern or understanding.)
+ Google’s search algorithm/Chatbot
Computer
vision for image and video processing
Speech
recognition for understanding natural language requests
Natural
language processing (NLP) for language understanding
- Backpropagation
+
Backpropagation (in finding errors and correcting) is a widely used method for
calculating derivatives inside deep feedforward (one-way) neural networks.
Backpropagation forms an important part of a number of supervised learning
algorithms for training feedforward neural networks, such as stochastic
gradient descent. OK, this is deep stuff, but, basically, experts have to train
the ANN to straighten-out it’s mistakes, like that just didn’t come out right! Yes,
an expert has to tune the ANN and correct it. I just can’t imagine a job like
this! Well, now we're going to look now at a massive workforce of humans whose
jobs were created by/for AI. They are the foot soldiers training the
algorithms. Their job title is sometimes called being an annotator or a tasker.
An AI annotator's role is to systematically review and label different data
types, translating human language and inputs into machine-understandable
formats. A Human-AI-Tasker typically refers to a system or framework where
humans (using AI Tasker Tools) and artificial intelligence collaborate to
complete tasks. This concept is often part of Human-Centered AI or
Humans-in-the-Loop systems, which emphasize the importance of human involvement
in AI processes. I know it, but don’t understand the mechanics of it! These are
some of the new work-anywhere jobs of AI, for qualified personnel. You have to
work smart and meet standards, but you can be paid electronically, daily $600
or more.
- Architectures and training
+ An AI Architect is a specialized
professional responsible for designing and overseeing the implementation of AI
solutions. Highly skilled, AI architects envision, build, deploy and
operationalize an end-to-end machine learning (ML) and AI pipeline. Now wait,
envision who this might be in your company! Watch out! Reality is, most officers,
executives, administrators, and managers, want AI, but history says, these
folks throw it in the closet, to the IT folks, and the customer service, business
risks, and security, somehow fall through the cracks. Yes, the company’s
employees and customers pay the price. Have you ever been the victim of a data
breach? This one is way worse!
Unit 7: Artificial intelligence in the 21st century
- Arriving at the current state of AI
+In
the 24th year of the 21st century, we now have AI, a
computer technology paradigm shift, that we must embrace and learn/employ our
critical thinking SKILLS. The experts agree that; AI development has flipped
over the last decade from academia-led to industry-led, by a large margin, and
this shows no sign of changing. Most companies can’t afford to develop it, so
they buy it from consulting companies. Consider Voice over IP, and telephone
answering system, the automated, digital corporate-telephone system. It asks
the customer to press a number or say “yes/no” to match your question, with a
VoiceIP answer or digital agent. To me this can be frustrating, and if you get
ugly, it hangs up on you. We often wonder if the company management ever
actually reviews their answering system. My general impression is that this is
NOT customer service, it’s customer aggravation and you’ll be hard-pressed to find
a human to complain too, because you’ll be holding for hours for a real-person.
What about monitoring and validating YOUR personal information that AI is
using? AI is different but, Will AI be
better or more frustrating?
- Big Data
+ Big data and AI are related but distinct
concepts. Big data is a collection of unstructured information from (wherever)
sources, while AI is a process of analyzing and learning from data. Big data is
the fuel that powers AI, providing it with the information necessary for
developing and improving features and pattern recognition capabilities. (data
is information and big information makes answers for you) AI, in turn, delivers
actionable insights (answers for you) from big data. Big data can come from the
Internet, publicly available sources, or it can be proprietary. I’m totally not
clear about how this is managed!
- Big Data and the Internet
+ Big data is produced from multiple data
sources like mobile apps, social media, emails, transactions or Internet of
Things (IoT) sensors, resulting in a continuous stream of varied digital
material. Not sure about the personal risks to private accounts, subscriptions,
or any personal stuff that we have. The Internet for sure is open season! For
years, I told users, students, and management that the Internet is the Wild
West; “connected”, you’re saying “here I am, try me”, and what you put out
there is fair-game for anyone, anywhere, for any reason, and it can stay there
forever! Great security is a tool but not a deterrence. Companies are now
experiencing that legislation or controls of Internet content are now working.
- AI and healthcare
+ This is a Big One for seniors; Artificial
intelligence in healthcare refers to the use of machine-learning algorithms and
software to mimic human cognition in the analysis, presentation, and
comprehension of complex medical and health care data. It involves the use of
machine learning, natural language processing, deep learning, and other
AI-enabled tools to assist and improve the patient experience, including
diagnosis, records, treatment, and outcomes. AI can help manage and analyze
data, make decisions, and conduct conversations, so it is destined to
drastically change clinicians’ roles and everyday practices. So many wearable
devices and personal monitors are already being incorporated into this technology.
This will be a big advantage to medical doctors and healthcare professionals,
and it WILL be the patients responsibility to monitor their personal data.
- AI and automobiles
+We
all know about this one and we accept it all, usually admitting that our cars
are smarter than us drivers. AI in vehicles has pros and cons. We would all
probably come up with the obvious but there’s more critical thinking needed.
- Cybersecurity
+ AI and cybersecurity:
AI is
used to automate repetitive tasks for security analysts.
It
helps identify shadow data and monitor data access.
AI
can anticipate cyberattacks and enable faster response.
Cybersecurity
is essential for the safety and reliability of AI systems.
How
does this help against data breach? Security tools are fantastic but they must
be judiciously managed, a real challenge for Information Management (Computer
people). Most of the big installations have a qualified data security manager.
- Machine translation
+ AI machine translation:
Uses
AI to automatically translate text and speech from one language to another.
Relies
on natural language processing and deep learning.
Aims
to preserve the meaning, context, and tone of the original content.
I use
Google Translate and Genie AI Chatbot, and look forward to more functions and
features.
- Ethics of AI
+ The ethics of artificial intelligence (AI)
is a critical field that addresses the moral and responsible development and
use of AI technology.
UNESCO’s
Recommendation on AI Ethics: In 2021, UNESCO produced the first-ever global
standard on AI ethics, known as the “Recommendation on the Ethics of Artificial
Intelligence.” This framework emphasizes four core values: respect for human
rights and dignity, promotion of diversity and inclusiveness, protection of the
environment and ecosystems, and ensuring transparency and accountability.
We
all want this, so given the history of computer technology and the Internet,
this will be a “bear” to achieve, and a never-ending struggle to maintain. Just
saying! My critical thinking says be really careful with your personal data AND
who you trust.
Unit 8: Data science and artificial intelligence
- What is data science?
+Data
Science is a process of collection and analysis of data. The data can come from
anywhere and anything. Every company and every computer user has data files and
the owner is responsible for protecting their data. Artificial Intelligence
(AI) involves the process of learning, reasoning, and self-correction from the
data. AI is limited to the implementation of machine learning algorithms,
whereas Data Science includes a broad range of statistical methods.
- Data science processes
+The
data science process is a structured approach to solving data-related problems.
Data science processes are a set of steps followed by data scientists as they
collect, analyze, model, and visualize large volumes of data. The process
covers everything from data collection to presenting visualized data and
insights to the business stakeholders.
Here
are the key steps involved:
1.Problem
Definition: Clearly define the problem and identify the goal of the analysis.
2.Data
Collection: Gather data from various sources. This can involve surveys, web
scraping, or accessing databases. This one is vague and suspicious but I really
don’t understand it all.
3.Data
Cleaning: Clean the data to remove duplicates, handle missing values, and
correct inconsistencies. This step ensures the data is ready for analysis.
4.Exploratory
Data Analysis (EDA): Analyze the data to uncover patterns, relationships, and
insights. This helps in understanding the data better and guides the modeling
process.
5.Model
Building: Use machine learning algorithms and statistical models to build
predictive models. This step involves selecting the right model, training it,
and evaluating its performance.
6.Model
Deployment: Deploy the model in a real-world environment where it can be used
to make predictions or provide insights. Monitoring the model’s performance is
crucial to ensure it continues to work well.
7.Communication:
Present the findings and insights to stakeholders in a clear and understandable
manner. This often involves visualizations and reports.
Now,
we can see that any data, anywhere can become the subject or object of AI
systems. Recently, Tik Toc, a popular video platform, used primarily by young
users, has become the topic of espionage risk. Kids, can’t understand the risk
of what they post on platforms nor can they understand becoming victims of the platform’s
sinister manipulation.
- Data exploration: An example
Data
exploration is the initial step in data analysis where you dive into a dataset
to get a feel for what it contains. It’s like detective work for your data,
where you uncover its characteristics, patterns, and potential problems. Data
exploration is an approach similar to initial data analysis, whereby a data
analyst uses visual exploration to understand what is in a dataset and the
characteristics of the data, rather than through traditional data management
systems. These characteristics can include size or amount of data, completeness
of the data, correctness of the data, possible relationships amongst data
elements or files/tables in the data. My first experience, of scraping, outside
of the private corporate data, was exploring into outside Geographic
Information Systems (GIS), to find stuff we could use. At the time, most GIS
systems used common software and data formats. It was kind of like; “if you
could get to it, it was free for the taking.”
Data
exploration is a hard-to-imagine AI job, but the tools are constantly
improving.
- AI methods in data science
+ AI methods are integral to data science,
enhancing the ability to extract meaningful insights from large datasets. Here is/are
some key AI methods used in data science:
1.Machine
Learning: This involves algorithms that learn from data to make predictions or
decisions without being explicitly programmed. Common techniques include
regression, classification, clustering, and reinforcement learning.
2.Deep
Learning: A subset of machine learning, deep learning uses neural networks with
many layers (hence “deep”) to model complex patterns in data. It’s particularly
effective in image and speech recognition. We’re seeing a lot of image stuff
now!
3.Natural
Language Processing (NLP): This method enables computers to understand,
interpret, and generate human language. Applications include sentiment
analysis, language translation, and chatbots.
4.Data
Mining: This involves exploring large datasets to discover patterns and
relationships. Techniques include association rule learning, anomaly detection,
and sequence mining.
5.Predictive
Analytics: Using historical data, predictive analytics employs statistical
algorithms and machine learning techniques to forecast future outcomes. It’s
widely used in finance, marketing, and healthcare.
6.Computer
Vision: This field focuses on enabling machines to interpret and make decisions
based on visual data from the world. Applications include facial recognition,
object detection, and autonomous vehicles.
- Autoencoders
+ An autoencoder is a type of neural network
architecture designed to efficiently compress (encode) input data down to its
essential features, then reconstruct (decode) the original input from this
compressed representation. A very complex subject but simply takes raw data and
converts it to fit your specific AI system.
- Data imputation
+ Data imputation is the process of replacing
missing or unavailable entries in a dataset with substituted values. This
process is crucial for maintaining the integrity of data analysis, as
incomplete data can lead to biased results and diminish the quality of the
dataset. From the early, beginning Information Management Systems, we inserted
default parameters to substitute for missing data and we created edit programs
that users could employ to research and complete, update, or correct the data
set, so this is a simple/primitive example.
Unit 9: Machine learning and artificial intelligence
- What is machine learning? (Two key aspects of
machine learning are big data(text,audio,images,video) and algorithms.)
+Machine
learning is another branch of artificial intelligence (AI) and computer
science. It focuses on the use of data and algorithms to imitate the way that
humans learn, gradually improving its accuracy. Machine learning allows a
computer program to learn and adapt to new data without human intervention. A
complex algorithm or source code is built into a computer that allows for the
machine to identify data, in a neural network, and build predictions around the
data that it identifies.
- Supervised learning
+Supervised
learning, also known as supervised machine learning, is a subcategory of
machine learning and artificial intelligence. It’s defined by its use of
labeled data sets to train algorithms to classify data or predict outcomes
accurately. Say what?.. Well, it’s like you ask AI (Input) and it works through
the Input/Hidden/Output layers of as neural network, using a compilation
match-stuff to answer your question (Output), AND, it can additionally find or
learn new possible matches.
+The
following are some of the common steps involved in supervised learning:
(Supervised
learning involves a human “teacher” or “supervisor.” This is becoming a popular
and lucrative new job description for work-from-anywhere and get paid daily.)
Gather
labeled data
1.Divide
the data into two sets: Training and Testing
2.Select
an appropriate algorithm (AI is a whole bunch of algorithms)
3.On
the training set, train the algorithm
4.Analyze
the algorithm’s performance on the testing set
5.If
necessary, fine-tune the model to improve performance
6.Make
predictions on new, unlabeled data using the trained model
- Unsupervised learning
+Unsupervised
learning in AI refers to: (It doesn’t need anyone to supervise the model.)
A
deep learning technique that identifies hidden patterns or clusters in raw,
unlabeled data. A type of machine learning where the algorithm is given input
data without explicit instructions on what to do with it. The training of a
machine using information that is neither classified nor labeled, allowing the
algorithm to act on that information without guidance. Creating a model that
extracts patterns from unlabeled data, (picking and placing) without
pre-existing labels.
Supervised vs Supervised
Learning: Supervised and Unsupervised learning
are the two techniques of machine learning. But both the techniques are used in
different scenarios and with different datasets.
- In
supervised learning, the AI is trained using a labeled dataset, which means
that each training example comes with the correct answer (label).
-The
goal is to learn a mapping from inputs to outputs, so the AI can predict the
label for new, unseen data.
- So,
Imagine teaching a child to recognize fruits. You show them pictures of apples
and bananas along with the labels "apple" and "banana."
After enough examples, the child can identify a new fruit based on what they’ve
learned.
- In
unsupervised learning, the AI is trained using an unlabeled dataset, meaning
there are no correct answers provided. The AI tries to learn patterns and
structures from the data on its own.
-The
goal is to find hidden patterns or groupings in the data.
-
Think of a child sorting a pile of mixed fruits without knowing what they are.
The child might group similar-looking fruits together, like all the round ones
in one pile and the long ones in another, even if they don’t know the specific
types.
Summary:
-Supervised
Learning: Learns from labeled data to make predictions.
-Unsupervised
Learning: Learns from unlabeled data to find patterns or groupings.
In
short, supervised learning needs answers to learn, while unsupervised learning
explores the data without any answers!
So, A
supervised learning model learns to classify data or accurately predict unseen
data based on labeled examples. In contrast, unsupervised learning aims to
discover hidden patterns, groupings, and dependencies within unlabeled data and
leverages it to predict outcomes. Unsupervised learning is used to convert
unlabeled data to labeled data, based on the patterns in the unlabeled data. So
that’s about as clear as mud, but it’s real.
- Reinforcement Learning
+ Reinforcement
learning is an autonomous, self-teaching system that essentially learns by
trial and error. It performs actions with the aim of maximizing rewards, or in
other words, it is learning by doing, in order to achieve the best outcomes.
Reinforcement
learning differs from supervised learning in a way that in supervised learning
the training data has the answer key with it so the model is trained with the
correct answer itself, whereas in reinforcement learning, there is no answer
but the reinforcement agent decides what to do to perform the given task. In
the absence of a training dataset, it is bound to learn from its experience.
Just more confusion but nice to know the general description!
Unit 10: Testing artificial intelligence systems
+I’m real concerned that planning, design, development,
testing, documentation, training, support, and system documentation/maintenance
were not more emphasized in this syllabus, because of its critical nature.
- Why test AI systems?
+As
AI continues to revolutionize industries, the role of testing becomes
increasingly vital. By implementing robust testing strategies and embracing
best practices, organizations can unleash the full potential of AI while
ensuring its reliability, security, and ethical use. The journey into the era
of AI is an exciting one, and with meticulous testing, we can navigate this
frontier with confidence, unlocking unprecedented opportunities for innovation
and advancement. Embrace the power of AI, test diligently, and pave the way for
a future where intelligent systems redefine what’s possible.
Testing
for AI systems comes with unique challenges, and requires specialized
techniques:
1.The
results of these AI-based systems are non-deterministic, i.e., they generate
different results for the same input.
2.There
is usually human bias in the training and testing data, which needs to be
identified and eliminated during AI model testing.
3.AI
performs best when given advanced input models.
4.AI
is an intricate system, and even small defects are magnified significantly.
- Software lifecycle costs
+The
lifecycle costs of AI software can vary significantly depending on the
complexity and scope of the project. Here’s a breakdown of the typical costs
involved:
1.Requirement
Analysis and Design: This initial phase involves extensive consultation and
requirement analysis sessions to conceptualize the AI system’s functionalities
and user interface design.
2.Development
and Testing: This phase includes the actual coding, integration, and rigorous
testing of the AI software.
3.Deployment
and Maintenance: Once the AI software is developed, it needs to be deployed and
maintained. This includes regular updates, bug fixes, and performance
monitoring.
4.Operational
Costs: These include the costs of running the AI software, such as cloud
computing resources, data storage, and energy consumption. These costs can be
substantial, especially for large-scale AI applications.
5.Training
and Support: Training users and providing ongoing support is crucial for the
successful adoption of AI software. This can involve additional costs depending
on the level of support required.
6.Lifecycle
Management: Managing the AI lifecycle involves continuous monitoring, updating,
and optimizing the AI models to ensure they remain effective and relevant. This
necessary function adds to the overall costs.
Most
companies will purchase AI solutions from vendors. Have you ever worked where
this was the case? It takes more than most companies want to do but they do it
because it’s an economic/efficiency opportunity! Employees and customers
suffer, first, from the poor planning and risk management decisions.
- Increasing adoption of AI/ML
+ The adoption of artificial intelligence
(AI) and machine learning (ML) is rapidly increasing across business sectors,
driven by their potential to revolutionize industries and improve efficiency.
Here are some key trends and impacts:
Technological
Advancements:
The
integration of big data and cloud computing with AI/ML is enabling more
effective deployment and scalability of these technologies.
Impacts
of AI/ML Adoption
1.Operational
Efficiency:
AI/ML
can automate routine tasks, allowing human workers to focus on more complex and
creative activities. This leads to increased productivity and efficiency across
various sectors.
2.Enhanced
Decision-Making:
By
analyzing large datasets, AI/ML can provide insights that help organizations
make more informed decisions, leading to better outcomes and competitive
advantages.
3.Security
Improvements:
AI/ML
technologies are being used to enhance cybersecurity measures, detect
anomalies, and prevent potential threats in real-time34.
4.Economic
Growth:
The
widespread adoption of AI/ML is expected to contribute significantly to
economic growth by creating new job opportunities and driving innovation.
Challenges
and Considerations
1.Ethical
and Privacy Concerns:
The
use of AI/ML raises important ethical questions, particularly around data
privacy and bias in algorithms. Ensuring transparency and fairness in AI/ML
applications is crucial.
2.Skill
Gaps:
There
is a growing demand for skilled professionals who can develop and manage AI/ML
systems. Addressing this skill gap through education and training is essential
for continued growth.
3.Regulatory
Frameworks:
Developing
robust regulatory frameworks to govern the use of AI/ML is necessary to ensure
these technologies are used responsibly and ethically. The increasing adoption
of AI/ML is transforming industries and driving innovation, but it also
requires careful consideration of ethical, privacy, and regulatory issues to
maximize its benefits.
- Uncertainty and Oracles
+ Uncertainty in AI refers to the inherent
unpredictability and ambiguity in data and decision-making processes. This can
arise from various sources, such as noisy or incomplete data, model
limitations, and the complexity of real-world environments. Managing this
uncertainty is crucial for developing robust AI systems.
+Oracles
in AI are theoretical entities or mechanisms used to provide correct answers or
guidance during the training and evaluation of AI models. They help in
validating the performance of AI systems by offering a benchmark or ground
truth against which the AI’s predictions can be compared
- Ethical dilemmas
+ Ethical dilemmas in AI include:
1.Bias:
AI algorithms and training data may contain biases.
2.Data
privacy and protection: Concerns about privacy and surveillance.
3.Decision
accountability: The role of human judgment in AI.
4.Environmental
impact: Considerations related to AI's carbon footprint.
5.Effects
on the workforce: Unemployment and income inequality due to automation.
- Adversarial inputs
+ Adversarial inputs are specially crafted
inputs that have been developed with the aim of being reliably misclassified in
order to evade detection. They are created to mislead machine learning models
into making inaccurate and wrong predictions. Adversarial examples are inputs
to machine learning models that an attacker has intentionally designed to cause
the model to make a mistake Vendors and defending organizations can defend
against malicious Adversarial Machine Learning by generating many adversarial
examples and then training the ML model to manage them properly. OK. No company
wants this to happen, so who you gonna call? Now, with 40 years in computer
technology, I saw 2 types of technicians:
1. Some fight change.
2. Others embrace it.
Only embracers will be
up to this challenge!
- Challenges of testing AI
+With
40 years in computer technology, issues in planning, design, development,
testing, training, documentation, and support of computer technology have
always been a problem. Smart folks don’t like mundane, repetitive stuff and
rebuff it’s importance. This human characteristic will never change and must be
constantly managed. So, this is a good AI application!
+ Challenges in AI testing seem to be
primarily on AI vendors, but also on large companies, developing custom AI
models, so challenges will include:
1.Complexity
of AI models
2.Lack
of standard testing frameworks
3.Data
quality and bias
4.Interpreting
AI decisions
5.Integrating
AI testing into development lifecycle
- Testing in machine learning
+ Machine learning testing involves
evaluating and validating the performance of ML models to ensure correctness,
accuracy, and robustness. Unlike traditional software testing, ML testing
includes additional layers due to the complexity of ML models1. There are four
major types of tests used in ML development:
1.Unit
tests: on individual components with single responsibilities.
2.Integration
tests: on combined functionality of individual components.
3.System
tests: on the design of a system for expected outputs given inputs
4.In
a controlled test-bed, use employees and actual test cases to prove the
validity of the AI system
- AI within wider systems
+Harvard
Business News nailed it; “Design AI systems for humans, by humans. The leading
companies in our research recognize that AI now allows them to build systems
that talk, listen, see, and understand much the way we do. They know that
tomorrow’s advantage will go to those who design systems that adjust to people
— not those who continue to expect people to adjust to systems.”
CUSTOMER
SERVICE has been losing to system requirements! If you’re like me, calling to
get a digital answering service that, after 1-9 options, you’re not finding an
option to meet your need, then holding for an inordinate time, only to get
transferred to several departments and repeatedly, stating your question, finding
only more confusion, AI must alleviate this situation!
