Autonomy in targeting is a function that could be applied to any intelligent system, in particular the rapidly expanding array of robotic systems, in the air, on land and at sea – including swarms of small robots. This is an area of significant investment and emphasis for many armed forces, and the question is not so much whether we will see more intelligent robots, but whether and by what means they will remain under human control. Today’s remote-controlled weapons could become tomorrow’s autonomous weapons with just a software upgrade. The central element of any future autonomous weapon system will be the software. Military powers are investing in AI for a wide range of applications10 and significant efforts are already underway to harness developments in image, facial and behavior recognition using AI and machine learning techniques for intelligence gathering and “automatic target recognition” to identify people, objects or patterns. Although not all autonomous weapon systems incorporate AI and machine learning, this software could form the basis of future autonomous weapon systems.
[DSC Europe 22] On the Aspects of Artificial Intelligence and Robotic Autonomy - Dr Pushan Kumar Dutta
1. On the Aspects of Artificial
Intelligence and Robotic Autonomy
2. Outline
01
• Hype vs reality
• General Intelligence vs Narrow
Intelligence
• What AI can and cannot do
• Machine Learning & Deep
Learning
• How does it work?
What is AI?
02
• What makes an AI company
• Virtuous cycle of AI
• Data Maturity
• Five steps of AI transformation
AI Transformation
03
• Marketing Analytics
• Automation
• Personalization
• CDP
AI in Marketing
04
• Ethical Issues
• Explainability
• Economical impact
• Limitation
AI and Society
4. AI Definitions
• The study of how to make programs/computers do
things that people do better
• The study of how to make computers solve problems
which require knowledge and intelligence
• The exciting new effort to make computers think …
machines with minds
• The automation of activities that we associate with
human thinking (e.g., decision-making, learning…)
• The art of creating machines that perform functions
that require intelligence when performed by people
• The study of mental faculties through the use of
computational models
• A field of study that seeks to explain and emulate
intelligent behavior in terms of computational
processes
• The branch of computer science that is concerned
with the automation of intelligent behavior
Thinking
machines or
machine
intelligence
Studying
cognitive
faculties
Problem
Solving and
CS
5.
6. The 1960s
• AI attempts to move beyond toy domains
• Syntactic knowledge alone does not work, domain
knowledge required
• Early machine translation could translate English to Russian
(“the spirit is willing but the flesh is weak” becomes “the
vodka is good but the meat is spoiled”)
• Earliest expert system created: Dendral
• Perceptron research comes to a grinding halt when it is
proved that a perceptron cannot learn the XOR operator
• US sponsored research into AI targets specific areas –
not including machine translation
• Weizenbaum creates Eliza to demonstrate the futility of
AI
7. 1970s
• AI researchers address real-world problems and solutions through
expert (knowledge-based) systems
• Medical diagnosis
• Speech recognition
• Planning
• Design
• Uncertainty handling implemented
• Fuzzy logic
• Certainty factors
• Bayesian probabilities
• AI begins to get noticed due to these successes
• AI research increased
• AI labs sprouting up everywhere
• AI shells (tools) created
• AI machines available for Lisp programming
• Criticism: AI systems are too brittle, AI systems take too much time
and effort to create, AI systems do not learn
8. 1980s: AI Winter
• Funding dries up leading to the AI Winter
• Too many expectations were not met
• Expert systems took too long to develop, too much money to
invest, the results did not pay off
• Neural Networks to the rescue!
• Expert systems took programming, and took dozens of man-
years of efforts to develop, but if we could get the computer
to learn how to solve the problem…
• Multi-layered back-propagation networks got around the
problems of perceptrons
• Neural network research heavily funded because it promised
to solve the problems that symbolic AI could not
• By 1990, funding for neural network research was
slowly disappearing as well
• Neural networks had their own problems and largely could
not solve a majority of the AI problems being investigated
• Panic! How can AI continue without funding?
9. 1990s: ALife
• The dumbest smart thing you can do is staying alive
• We start over – lets not create intelligence, lets just create
“life” and slowly build towards intelligence
• Alife is the lower bound of AI
• Alife includes
• evolutionary learning techniques (genetic algorithms)
• artificial neural networks for additional forms of learning
• perception and motor control
• adaptive systems
• modeling the environment
• Let’s disguise AI as something new, maybe we’ll get
some funding that way!
• Problems: genetic algorithms are useful in solving some
optimization problems and some search-based problems, but
not very useful for expert problems
• perceptual problems are among the most difficult being
solved, very slow progress
10. Today: The New (Old) AI
• Look around, who is doing AI research?
• By their own admission, AI researchers are not doing “AI”, they are
doing
• Intelligent agents, multi-agent systems/collaboration
• Ontologies
• Machine learning and data mining
• Adaptive and perceptual systems
• Robotics, path planning
• Search engines, filtering, recommendation systems
• Areas of current research interest:
• NLU/Information Retrieval, Speech Recognition
• Planning/Design, Diagnosis/Interpretation
• Sensor Interpretation, Perception, Visual Understanding
• Robotics
• Approaches
• Knowledge-based
• Ontologies
• Probabilistic (HMM, Bayesian Nets)
• Neural Networks, Fuzzy Logic, Genetic Algorithms
11. AI can help Small and Medium-sized Enterprises (SMEs) do their tasks in more efficient ways, making them more
productive, looking forward to which SMEs are embracing AI.
Why SMEs are important?
Because SMEs are primarily responsible for economic growth and prosperity. Their capacity for innovation and
flexibility in a changing business environment makes them crucial for success in the global economy.
SMEs also charge less to their clients than some of the big companies and this also allows the clients to benefit from
cheaper and hence more profitable deals as they attract their way out for clients and provide special offers with
reduced prices to increase their business or for the stable position in the market.
In fact, it is these SMEs that governments rely upon to increase growth in their economies. The biggest asset of SMEs
is that it provides employment.
Should SMEs embrace Artificial Intelligence?
Yes, they should, as these tools can help small businesses to thrive in a changing marketplace.
12.
13. Key Steps in Machine Learning
1. Collect data
2. Train model
Iterate many times
3. Deploy model
Get data back
Update/maintain model
Key Steps in Data Science
1. Collect data
2. Analyze Data
3. Suggestion
Machine Learning vs Data Science
15. Input (A) Output (B) Application
Email Spam? (0/1) Spam filtering
Audio file Text transcripts Speech recognition
English Indonesian Machine translation
Ad, user info Click? (0/1) Online advertising
House features House price Real estate
Image, radar info Position of other
cars/objects
Self-driving car
Image of product Defect? (0/1) Visual inspection
Supervised Machine Learning
Supervised
Learning
A
(input)
B
(output)
21. VOICE AND FACE
RECOGNITION
TARGETED
ADVERTISING AND
REMARKETING
CHATBOTS, ONLINE
SUPPORT
AND VIRTUAL
ASSISTANCE
AGENTS
PREDICTIVE
ANALYTICS AND
CUSTOMER
SERVICE
WEB SEARCH
SUGGESTIONS
EMAIL SPAM
FILTERS AND
CATEGORIZATION
AUTOMATIC
SCHEDULING
INTERNET OF
THINGS
Artificial intelligence has helped research and develop the following aspects of the business
that companies use for their financial and market positioning advancement:
22. Intelligent Robots Data in Digital Transformation
•Customer service: Amazon has deployed neural networks to generate personalized product
recommendations for customers, bridging the gap between their huge product catalog and sparse datasets
for each individual customer, as a result of the small amount of products any individual typically
purchases;[4]
•
Medical diagnostics: A CNN, or deep convolutional neural networks – an AI system based on neural
networks, is capable of classifying skin cancer with a level of competence allegedly comparable to
dermatologists. It can be run on a smartphone, therefore potentially providing universal access to low-cost
diagnostic advice.[5]
•
Cyber security: AI2, developed at MIT’s Computer Science and Artificial Intelligence Laboratory, scans and
reviews tens of millions of log lines each day and pinpoints anything suspicious to be escalated to a human
being. AI2 successfully identifies 86% of attacks whilst sparing analysts the time and effort of following up
on false alarms.
23. 5 Ways Big Data Promoted AI Implementation:
Increased processing capability: With the evolution of processors in the recent years, there has been drastic growth in
computing speeds. Billions of instructions can be processed in a few microseconds.
Availability of low cost and large scale memory device: High storage and retrieval of big data is now possible using
efficient memory devices like DRAM’s (Dynamic Random Access Memory) and logic gate such as NAND’s.
Learning from actual data sets and not from sample ones: Big data analysis supports the breakdown of data sets to
identify words and phrases. Similar is the case with image processing, as it identifies appearances, outlines, and maps to
process information. Big data analysis enables machines to recognize images and learn how to respond.
Open-source programming languages and platforms:Hadoop is an open source, java based software framework that
has capabilities of reading and analyzing distributed data sets. Since Hadoop is open source, it is reliable and a free
programming tool for data analysis. It has made AI algorithm execution more efficient.
Robotic platforms to assist isolated patients
Hospitalized patients in isolation do not have the possibility to receive visits or be accompanied, due to the strict
restrictions to stop the spreading of the virus. Moreover, in many cases, the overload of healthcare systems reduces the
time that clinical staff can dedicate to each patient. Therefore, they spend most of their time alone which may further
cause physical and psychological deterioration.
Robotic platforms can disinfectation
As recently seen in China, autonomous robotic platforms for disinfection of healthcare facilities can play a
fundamental role, by preventing and reducing the spread of infectious diseases, virus, bacteria and other types of
harmful organic microorganisms in the environment. They are particularly effective when equipped with ultraViolet
(UV) light emitters which break down the DNA-structure of viruses. In other cases, robots and drones have been used
to spray sanitizers in public spaces to mitigate the possible expansion of the infection.
24. AI & Finance
1.Risk assessment
Can you use artificial intelligence to determine whether someone is eligible for a loan? Definitely. In fact, banks
and apps are using machine learning algorithms to not only determine a person’s loan eligibility
2.Risk management
identify risks, conserve manpower and ensure better information for future planning,”.
3.Fraud detection, management and prevention
Have you ever received a phone call from your credit card company after you’ve made several purchases?
4.Credit decisions
Assess a potential customer based on a variety of factors, including smartphone data (plus, machines aren’t
biased.)
5.Financial advisory services
Looking to follow the latest financial trends? Interested in a portfolio review?
6.Trading
Since artificial intelligence is used to analyze patterns within large data sets, it’s no surprise that it’s often used in
trading. Managing finances/personalized banking
Chatbots and virtual assistants have reduced (and in some cases eliminated) the need to spend time on the
phone.
7.Preventing cyberattacks
Consumers want to be reassured that banks and financial institutions will keep their money and personal
information
8.Better predict and assess loan risks
25. Security and
Intelligent
applications
Intelligent
Mobility Systems
AI and Autonomy
in Space
Big Data and
Mobile
ecommerce
Quantum
Computing + AI
Neural Science
HRI and social
robotics
Robot business,
marketing &
economic impact
Rehabilitation
and transfer
robots
Robotics and
autonomous
driving
Bio-inspired
robotics
Applied
telerobotics
Robotic Mapping
Biomedical
Robotics
Emerging Smart
Materials
Optimization
Algorithms
Computer-assisted
Navigation
Architectural
Robotics
Robot mechanics,
dynamics and
motion control,
Space robot, flexible
robot, service robot
Surgical robot theory
and application
Parallel robot and
parallel machine tool
Mobile robot control
and navigation
Homecare Robotics,
Rescue Robotics,
Cooperative
Robotics
Orthotics,
Prosthetics,
Wearable Robotics
Systems science
and Enginering
26. When AI is about to be implemented into business practices and procedures, this can be optimized so that your company makes
most of its full potential.
What follows are different methods that can be applied to taking full advantage of AI processes and systems within your
organization.
Where Should You Implement AI?
It goes without saying that you should first figure out what is it that your business needs by determining what issues can be
eliminated or points improved using AI.
Automatization of tasks, prioritizing a concrete value (appraisal of the business and financial price of AI implementation), adding
AI capacities to your existing products and services
By using AI’s smart algorithms and instant profiling, your company can display pertinent advertisement campaigns and singular
ads to people that are most likely to be your potential buyers.
These smart algorithms can be applied in social media marketing, email marketing or paid search. This method cuts down
expenses from serving ineffective ads (to wrong audiences). Instead of that, they can be used to communicate with the target
audience in a fashion that increases their engagement.
As business operations increase and platforms used for business amass, a company will find itself in a need of AI software for
management and consolidation of all the data gathered from these resources into a single dashboard.
27. 1) All Artificial Popeyes are initialized
2) Generations count is initialized with zero
3) All Artificial Popeyes identify their local best
4) All Artificial Popeyes identify their global best
5) loop through for each particle do 6) if (
random_number_generate (0,1) < PopeyeSpinachProbability
) then
7) Popeye with Spinach updates Velocity
8) Popeye with Spinach updates Position 9) else // Popeye
without Spinach
10) if ( random_number_generate (0,1) <
HelpOfPopeyeWithSpinachProbability) then
11) Popeye without Spinach updates Velocity
12) Popeye without Spinach updates Position
13) else
14)
15) end if
16) end if
17) end for
18) generations += 1
28. 1) All Artificial Chhota Bheems are initialized
2) Generations count is initialized with zero
3) All Artificial Chhota Bheems identify their local best
4) All Artificial Chhota Bheems identify their global
best
4) loop through for each particle do
5) if ( random_number_generate (0,1) <
ChhotaBheemLadduProbability ) then
7) Chhota Bheem with Laddu updates Velocity
8) Chhota Bheem with Laddu updates Position
9) else // Chhota Bheem without Laddu
10) if ( random_number_generate (0,1) <
HelpOfChhotaBheemWithLadduProbability) then
11) Chhota Bheem without Laddu updates Velocity
12) Chhota Bheem without Laddu updates Position
13) else 14)
15) End
if 16) end
if 17) end for
18) generations += 1
19) while (termination condition not reached)
29. 1) All Artificial Jerrys are initialized
2) Generations count is initialized with zero
3) All Artificial Jerrys identify their local best
4) All Artificial Jerrys identify their global best
5) loop through for each particle do
6) if ( random_number_generate (0,1) <
JerryCheeseProbability ) then
7) Jerry with Cheese updates Velocity
8) Jerry with Cheese updates Position
9) else // Jerry without Cheese
10) if ( random_number_generate (0,1) <
HelpOfJerryWithCheeseProbability) then
11)Jerry without Cheese updates Velocity
12) Jerry without Cheese updates Position
13)else 14)
14) end if 16) end if
15) end for
16) generations += 1
17) while (termination condition not reached)
30. 1) All Artificial Happy Kids are initialized
2) Generations count is initialized with zero
3) All Artificial Happy Kids identify their local best
4) All Artificial Happy Kids identify their global best
5) loop through for each particle do
6) if ( random_number_generate (0,1) <
HappyKidBananaProbability ) then
7) Happy Kid with Banana updates Velocity
8) Happy Kid with Banana updates Position
9) else //
10)if ( random_number_generate (0,1) <
HelpOfHappyKidWithBananaProbability) then
11)Happy Kid without Banana updates Velocity
12)Happy Kid without Banana updates Position
13)Else
14) 15) end if
15) end if
16) end for
17) generations += 1
18)while (termination condition not reached)
31. There are no Hybrid PSO algorithms based on Robot Selection till date. There
are 2 options to select from for Robots . Either Robots move towards local best
position or they move towards global best position. Based on random number
generated and Robot Selection Probability, Robot select from 2 options
available. If random number generated is less than Robot Selection Probability
then Robot move towards local best as shown in line number 15. Otherwise,
Robot move towards global best position as shown in line number 20.
34. 5 Steps of AI Transformation
1. Execute pilot projects to gain momentum
2. Build an in-house AI team
3. Provide broad AI training
4. Develop an AI strategy
5. Develop internal and external communications
Source: https://landing.ai/ai-transformation-playbook/
36. • Explainability of AI is hard.
• Ethical issues.
• Privacy compliance.
• AI can become biased with biased data.
• AI systems are open to Adversarial Attacks. In future companies
might be at war with the adversarial attackers.
• US and China are leading in AI but this technology is still
immature giving other nations an equal advantage to compete.
• AI to create 13 Trillion Value by 2030 mostly to be used in Retail
followed by Travel and Automotive sector. (McKinsey)
• By 2030 (McKinsey):
• Jobs displaced by AI : 400–800 Million
• Jobs created by AI : 555–890 Million
Some notes on AI