Explainable AI (xAI) & Future of Artificial Intelligence/ML (HLAI & AGI)
1st/old version presented @ CODECON 2023 #CODECON, #CODECON2023
Radovan Kavický, AIslovakIA & GapData Institute
21. 2. 2024
(#AIslovakIA, #AIRadioMeetup, Bratislava, Slovakia) #AIslovakIA @ sli.do�
Join at slido.com�#AIslovakIA
ⓘ Start presenting to display the joining instructions on this slide.
#AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Structure of the talk
- 1. Current development and what’s next/what to do now? [15 min.]
Sora, Gemini 1.5, Copilot/ChatGPT & GitHub Copilot
- 2. Future of AI (path towards HLAI/AGI, when? why?) [20 min.]
Strong/Wide AI vs. Weak/Narrow AI, Machine Learning vs. Machine Reasoning
- 3. Ethics & Transparency in AI/Trustworthy AI/Responsible AI/Applied AI/Growth potential [10 min.]
Brain/Neuroscience Research and theory of consciousness
- 4. Explainable AI/xAI, Shapley values, Mathematics of AI [20 min.]
- 5. Conclusion & position of Slovakia/Europe [10 min. + Q&A/15-20 min]
—--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
- Data Science (definition): collection of scientific results and methods for transformation of data from raw form to meaningful information, knowledge and wisdom, which should support better decisions
- AI (definition) - theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, translation and others
1/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Which programming language do you prefer for AI? (Python, C, Scala, R, Julia, MATLAB, whatever)
ⓘ Start presenting to display the poll results on this slide.
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
2/12 #AIslovakIA @ sli.do
About me
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
3/24 #AIslovakIA @ sli.do
What IDE you are using for your AI projects/where you create/train your models? (cloud, locally, etc.)
ⓘ Start presenting to display the poll results on this slide.
4/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
AI, AGI, HLAI > Palantir, Flawless, Sora and beyond… Frankie.AI goes to Hollywood
- Complex, modern-day ML algorithms, where deep learning & ensemble methods dominate
- really hard to fully understand, but the decision process behind them
- ChatGPT is not transparent/open (at all/bans and restrictions)
- can & need to be transparent and trustworthy for decision makers
- AI & BI coming together
- critical domains as finance, healthcare or public sector & governmental services, where TRUST is a MUST
- Explainability vs. Simplicity
- growing regulatory pressure also outside these areas Explainable AI (xAI) will be necessity for any organization soon
- understand the inner workings of these ML algorithms & how to design systems that imitate intelligence in a transparent way
- overview of current trends in Explainable AI/ML & the challenges that are ahead of us
- Next big thing?
5/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Global GDP ≈ $100 Trillion USD
6/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
$100,562,000,000,000 (2022, nominal/per year)
US: 25.5 Trillion USD/year (GDP)
China: 17.9 Trillion USD/year (GDP)
Japan: 4.2 Trillion USD/year (GDP)
Germany: 4 Trillion USD/year (GDP)
India: 3.4 Trillion USD/year (GDP) +UK, France,
Russia, Canada, Italy (about 2 Trillion USD/year)
Explainable Machine Learning/AI (xAI), Era of Chatbots, AI & BI collide
- Complex, modern-day ML algorithms, where deep learning & ensemble methods dominate
- really hard to fully understand, but the decision process behind them
- ChatGPT is not transparent/open (at all/bans and restrictions)
- can & need to be transparent and trustworthy for decision makers
- AI & BI coming together
- critical domains as finance, healthcare or public sector & governmental services, where TRUST is a MUST
- Explainability vs. Simplicity
- growing regulatory pressure also outside these areas Explainable AI (xAI) will be necessity for any organization soon
- understand the inner workings of these ML algorithms & how to design systems that imitate intelligence in a transparent way
- overview of current trends in Explainable AI/ML & the challenges that are ahead of us
- Next big thing?
7/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Source: Andrew Ng, deeplearning.ai
Explainable Machine Learning/AI (xAI), Era of Chatbots, AI & BI collide
- Complex, modern-day ML algorithms, where deep learning & ensemble methods dominate
- really hard to fully understand, but the decision process behind them
- ChatGPT is not transparent/open (at all/bans and restrictions)
- can & need to be transparent and trustworthy for decision makers
- AI & BI coming together
- critical domains as finance, healthcare or public sector & governmental services, where TRUST is a MUST
- Explainability vs. Simplicity
- growing regulatory pressure also outside these areas Explainable AI (xAI) will be necessity for any organization soon
- understand the inner workings of these ML algorithms & how to design systems that imitate intelligence in a transparent way
- overview of current trends in Explainable AI/ML & the challenges that are ahead of us
- Next big thing?
8/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Source: HLAI Conference | GoodAI https://www.goodai.com/hlai/ Ben Goertzel, 2018 (Good.AI & GapData Institute, PyData Slovakia/partnership)
Explainable Machine Learning/AI (xAI), Era of Chatbots, AI & BI collide
- Complex, modern-day ML algorithms, where deep learning & ensemble methods dominate
- really hard to fully understand, but the decision process behind them
- ChatGPT is not transparent/open (at all/bans and restrictions)
- can & need to be transparent and trustworthy for decision makers
- AI & BI coming together
- critical domains as finance, healthcare or public sector & governmental services, where TRUST is a MUST
- Explainability vs. Simplicity
- growing regulatory pressure also outside these areas Explainable AI (xAI) will be necessity for any organization soon
- understand the inner workings of these ML algorithms & how to design systems that imitate intelligence in a transparent way
- overview of current trends in Explainable AI/ML & the challenges that are ahead of us
- Next big thing?
9/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Explainable Machine Learning/AI (xAI), Era of Chatbots, AI & BI collide
10/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
“The year is 2052 and the world is a dangerous and chaotic place. Terrorists operate openly - killing thousands; drugs, disease and pollution kill even more. The world's economies are close to collapse and the gap between the insanely wealthy and the desperately poor grows ever wider.”
Source: Warren Spector/Deus Ex (6/2000) + https://www.youtube.com/watch?v=tffX3VljTtI
Explainable Machine Learning/AI (xAI), Era of Chatbots, AI & BI collide
- Complex, modern-day ML algorithms, where deep learning & ensemble methods dominate
- really hard to fully understand, but the decision process behind them
- ChatGPT is not transparent/open (at all/bans and restrictions)
- can & need to be transparent and trustworthy for decision makers
- AI & BI coming together
- critical domains as finance, healthcare or public sector & governmental services, where TRUST is a MUST
- Explainability vs. Simplicity
- growing regulatory pressure also outside these areas Explainable AI (xAI) will be necessity for any organization soon
- understand the inner workings of these ML algorithms & how to design systems that imitate intelligence in a transparent way
- overview of current trends in Explainable AI/ML & the challenges that are ahead of us
- Next big thing?
5/12 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
AI & Ethics (food for thought & main challenges for the future)
- we should all be humble & think all the time about ethical implications of our work
- data in not (at least our) reality & need of transparency (explainable AI/HLAI/Superintelligence)
- we have no programming of common sense (fails/CYC)
- we need both left and right brain/lobe thinking
- we don't know how we really think (and if you think you know, well… good luck)
- we don't know what intelligence is (unknown processes)
- we are not walking neural nets (but if some of you are, please, let me know)
- design perceptron (imitate brain without knowing how the brain really works), real research of brain started (Brain Initiative/public and private research, 2013 +DARPA, IARPA)
- we need to rethink the basic design of neural nets (mathematics of deep learning)
- we can count all the neurons in our brains (10's of billions and 1000’s of billions synapses)
- we need rules (ethical & law standards/technology is currently before ethics)
“But the big feature of human-level intelligence is not what it does when it works but what it does when it's stuck.”
-- Marvin Minsky
11/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Explainable Machine Learning/AI (xAI), Era of Chatbots, AI & BI collide
- Complex, modern-day ML algorithms, where deep learning & ensemble methods dominate
- really hard to fully understand, but the decision process behind them
- ChatGPT is not transparent/open (at all/bans and restrictions)
- can & need to be transparent and trustworthy for decision makers
- AI & BI coming together
- critical domains as finance, healthcare or public sector & governmental services, where TRUST is a MUST
- Explainability vs. Simplicity
- growing regulatory pressure also outside these areas Explainable AI (xAI) will be necessity for any organization soon
- understand the inner workings of these ML algorithms & how to design systems that imitate intelligence in a transparent way
- overview of current trends in Explainable AI/ML & the challenges that are ahead of us
- Next big thing?
12/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Explainability & Tools (LIME, SHAP & others)
- LIME (Local Interpretable Model-agnostic Explanations)
- explainable AI method that helps to illuminate a machine learning model
- make its predictions individually comprehensible
- method explains the classifier for a specific single instance (suitable for local explanations)
- model agnostic tool which can give explanations for any supervised learning model
- one of the most popular XAI tools out there due to its simplicity and intuitiveness
https://github.com/marcotcr/lime
- SHAP (SHapley Additive exPlanations)
- game theoretic approach to explain the output of any machine learning model.
- connects optimal credit allocation with local explanations using the classic Shapley values from game theory and their related extensions
https://github.com/slundberg/shap
+
Others (100+ tools), f.e.:
- Alibi (Algorithms for explaining machine learning models)
https://github.com/SeldonIO/alibi
13/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Explainable AI/xAI + key literature
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
14/24 #AIslovakIA @ sli.do
AI & Slovakia or why should you even care?
+
+
15/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
AI & Slovakia, what exactly are you talking about? What future?
- OECD.AI that provides an overview of AI strategies and policies in Slovakia. It includes an action plan for the transformation of Slovakia into a successful digital country and development of the digital single market. The action plan offers a set of concrete steps on how to start building a sustainable and human centric, and trustworthy AI.
- AIslovakIA, which is a neutral, independent and non-profit platform with an ambition to develop excellence and bring together experts and those interested in artificial intelligence in Slovakia.
- AI/ML augmented analytics (scenarios for data-driven decision making)
- build and deploy highly accurate machine learning models without writing a single line of code, experiment, simulate and compare different scenarios using various models/scenarios to identify the best strategy or test ideas before committing resources
- focus on the right data to analyze, get predictive insights with explanations in your dashboards + run simulations to get actionable prescriptive guidance on what to do next and get instant visual response
- Alphaa.ai (world’s 1st AI voice analyst/not just regular “chat bot”)
16/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
- OECD.AI that provides an overview of AI strategies and policies in Slovakia. It includes an action plan for the transformation of Slovakia into a successful digital country and development of the digital single market. The action plan offers a set of concrete steps on how to start building a sustainable and human centric, and trustworthy AI.
- AIslovakIA, which is a neutral, independent and non-profit platform with an ambition to develop excellence and bring together experts and those interested in artificial intelligence in Slovakia.
- AI/ML augmented analytics (scenarios for data-driven decision making)
- build and deploy highly accurate machine learning models without writing a single line of code, experiment, simulate and compare different scenarios using various models/scenarios to identify the best strategy or test ideas before committing resources
- focus on the right data to analyze, get predictive insights with explanations in your dashboards + run
#AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Who's on the picture?
ⓘ
Click Present with Slido or install our Chrome extension to activate this poll while presenting.
17/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Future of AI/ML and Data Science
- Back to Basics
What Machine Learning/Artificial Intelligence is? What Data Science is?
“Artificial intelligence is the science of making machines do things that would require intelligence if done by men.”
-- Marvin Minsky
- Data Science (definition): collection of scientific results and methods for transformation of data from raw form to meaningful information, knowledge and wisdom, which should support better decisions
- Data Science: Statistics + programming; data analysis + Computer Science, modelling + Econometrics, Big Data, ML/DL
- Expectation/Prediction: Data Scientists, one of the first “victims” of automation
- Is Data Reality? Do we model reality? (noise/signal)
18/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
19/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
20/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Is data reality?
ⓘ Start presenting to display the poll results on this slide.
21/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Data is not reality. �And if you think so, you should quit doing so… right now.
Source: Me :), H2O.ai Meetup, Prague 2019, https://www.meetup.com/Prague-Artificial-Intelligence-Deep-Learning/events/264335458/
22/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
AI & Ethics of AGI (Transhumanism & Empathy/future challenges)
23/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Source: https://edition.cnn.com/2023/09/20/tech/musk-neuralink-human-trials/index.html
AI, Consciousness & Ethics of AGI (Transhumanism & Empathy/future challenges)
24/24 #AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
Future is awesome. �All we have to do now is to build it.
Source: PyData Berlin 2017 (w/Marek Rosa), Talk on YouTube: https://www.youtube.com/watch?v=I5578BhU4sE
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
#AIslovakIA @ sli.do
PyData Slovakia & activities within CEE + V4
#AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
GapData Institute (GDI) and how to support us.
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
#AIslovakIA @ sli.do
Thank you for your attention
Contact:
Radovan Kavicky
radovan.kavicky@aislovakia.com
radovan.kavicky@student.tuke.sk
+421 949 716 214 (SK)
+420 777 595 262 (CZ)
http://www.linkedin.com/in/radovankavicky
https://github.com/radovankavicky
https://www.facebook.com/groups/356635138031671
@radovankavicky, @PyDataBA, @GapDataInst
In case you have any question, feel free to ask.
#AIslovakIA @ sli.do
Oh, and one more thing… :)
https://www.meetup.com/Julia-Users-Group-Slovakia/
https://www.facebook.com/groups/379292635993253/
Hashtag: #JUGSlovakia
More to come... soon. Stay tuned.
#AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
How would you rate this talk? (scale 1-10, 10/awesome, 5/meh, 1/worst talk ever, but "try again, fail again, fail better")
ⓘ
Click Present with Slido or install our Chrome extension to activate this poll while presenting.
#AIslovakIA @ sli.do
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
(xAI) English Exam (translation/Slovak)
AI-2023 Forty-third SGAI International Conference on Artificial Intelligence
AI-2023 is the 43-rd Annual International Conference of the British Computer Society's Specialist Group on Artificial Intelligence (SGAI)
Alan Turing studied (math, 1931, Cambridge)
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
#AIslovakIA @ sli.do
(xAI) English Exam (translation/Slovak)
AI-2023 is the 43-rd Annual International Conference of the British Computer Society's Specialist Group on Artificial Intelligence (SGAI)
Alan Turing studied (math, 1931, Cambridge), Stephen Hawking lived few meters from Peterhouse College
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
#AIslovakIA @ sli.do
(xAI) English Exam (translation/Slovak)
AI-2023 is the 43-rd Annual International Conference of the British Computer Society's Specialist Group on Artificial Intelligence (SGAI)
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
#AIslovakIA @ sli.do
Audience Q&A Session
ⓘ Start presenting to display the audience questions on this slide.
Explainable Machine Learning/AI (xAI) & Future of Artificial Intelligence
#AIslovakIA @ sli.do