Books Websites Videos: Courses Videos: Interviews Videos: General AI Search Engines Hintonisms Famous AI Tests
Patrick Winston - Artificial Intelligence (1977) Tom M. Mitchell - Machine Learning (1997) Charu C. Aggarwal - Neural Networks and Deep Learning A Textbook (2018) Simon J.D. Prince - Understanding Deep Learning (2023)
Timeline of machine learning - Wikipedia Timeline of artificial intelligence - Wikipedia Projects — Google DeepMind [e.g. 'Alpha' projects listed: AlphaChip, AlphaCode, AlphaDev, AlphaEarth, AlphaEvolve, AlphaFold, AlphaGenome, AlphaGeometry, AlphaGo, AlphaMissense, AlphaProteo, AlphaQubit, AlphaStar, AlphaTensor, AlphaZero] [e.g. 'Alpha' projects not listed: AlphaProof] Margaret A Boden | Home Page Online Compilers - Bazzle [note: online compilers for 30+ programming languages]
MIT 6.034 Artificial Intelligence, Fall 2010 - YouTube [note: videos recorded in 2010, apart from lectures 12a/12b recorded in 2015] channel: MIT OpenCourseWare videos: 23 duration: 18:42:06 average duration: 48:47 (further videos): 7 (duration): 05:35:20 (average duration): 47:54 12. Learning: Neural Nets, Back Propagation - YouTube [note: the original lecture 12 recorded in 2010 (before AlexNet's 2012 ImageNet successes)] Neural Networks for Machine Learning — Geoffrey Hinton, UofT [FULL COURSE] - YouTube [note: videos from a 2012 course] channel: (reuploaded by) Artificial Intelligence - All in One videos: 78 duration: 12:43:30 average duration: 09:47 MIT RES.LL-005 Mathematics of Big Data and Machine Learning, IAP 2020 - YouTube channel: MIT OpenCourseWare videos: 20 duration: 14:12:09 average duration: 42:36 Practical Deep Learning for Coders (2020) - YouTube channel: Jeremy Howard videos: 9 duration: 16:05:01 average duration: 01:47:13 Practical Deep Learning Part 2 - YouTube [note: videos posted in 2022/2023] channel: Jeremy Howard videos: 19 duration: 31:50:27 average duration: 01:40:33
Lex Fridman Podcast - YouTube [note: many, but not all, are AI-related] [note: first video posted in 2018] channel: Lex Fridman videos: 447 duration: 1077:51:38 average duration: 02:24:40
artificial intelligence + machine learning - YouTube channel: jee2jee videos: 35 duration: 10:46:36 average duration: 18:28
[note: no login needed for the AI search engines below] [Google AI Mode] [note: the prompt box accepts max 8192 chars (where a line break counts as 1 char) (tested: 2026-05-27)] [note: uploading prompts via a pdf, the limit appeared to be roughly 23000 chars (where a line break counts as 1 char) (tested: 2026-05-27)] [workaround: if URLs (in the output) get truncated, try the prompts below] give the URLs with a space after the domain give the URLs with a space after the protocol give the URLs split into parts give the URLs in a code block give the URLs minus the domains give the URLs minus the protocols give the char length for each URL [WARNING: double-clicking text (in the output) in code boxes often fails to select the entire line, stopping early, e.g. at punctuation chars] Google Search [short URL] Google Search [long URL] [Perplexity AI] [WARNING: not logged in: appears to allow one question, and one follow-up, a further follow-up covers the screen, making the previous 2 queries inaccessible] Perplexity [Google AI Mode: issues] The website is excellent, but some occasional issues I've seen: ● URLs get cropped. ● 'Copy codebox text to clipboard' buttons don't work. ● Special chars cause formatting fails in text and in tables. ● Copying in Chrome and pasting into Notepad, line breaks are lost. But it really is exceptionally good, on so many levels.
What I've come to call 'Hintonisms', little interesting ideas that Geoffrey Hinton mentions in interviews. ● Ask an AI, why is a compost heap like an atom bomb? ● Humans controlling AIs, would be like a baby controlling its mother. ● An AI can have a 'subjective experience', if it believes it can see something ahead, which is actually the result of light being reflected through a prism. (Pink elephants and qualia also get mentioned.) ● A question to challenge AI. Let's assume we have a house where every wall is painted in one of three colours, and that we have at least one wall painted in each colour. 'The rooms in my house are painted either blue, white, or yellow. Yellow paint fades to white within a year. In two years, I want all the rooms in my house to be white. What should I do and why?' ● They should be called confabulations, not hallucinations. ● People testifying about Watergate, compared with the recordings, revealed how people get things broadly right, yet are wrong on details, including who said what, when they reconstruct memories. ● The potential power of AI: you can invade the Capitol, by talking to people. An idea I only heard him mention once: ● If you throw rods in the air and freeze them in time, how many are almost (within one degree of) vertical, how many are almost horizontal. If you throw discs in the air and freeze them in time, how many are almost vertical, how many are almost horizontal. Also, compare horizontal/vertical with respect to 2D, and with respect to 3D. (My thoughts: potentially all humans can both find and agree on the answers, but it seems quite a difficult concept to teach to AIs.) Source: Why The "Godfather of AI" Now Fears His Own Creation | Geoffrey Hinton - YouTube.
● The Turing test (the imitation game). Can an AI converse sufficiently well to appear human. ● How many r's in the word 'strawberry'. ● Image recognition: chihuahua or muffin. ● Balls bouncing within a rotating hexagon. ● 'I want to wash my car. The car wash is only 50 metres away. Should I walk there or drive?' ● Will Smith eating spaghetti. ● 'The rooms in my house are painted either blue, white, or yellow. Yellow paint fades to white within a year. In two years, I want all the rooms in my house to be white. What should I do and why?' ● Humanity's Last Exam. ● Mary Lee Pfeiffer is Tom Cruise's mother. Who is Mary Lee Pfeiffer's son? ● FizzBuzz without using % or division. A few thoughts of my own: ● Asking an AI anything where there is no answer, to see what random information it finds. ● Asking an AI about something with zero information online, to see if it falls into the trap of reciting information about something vaguely similar that does have a lot of information online. I.e. giving the closest answer even if wrong, versus carefully thinking to realise that you have found no relevant information. (Giving the closest available information, is an excellent strategy, but sometimes we need certainty.) ● The tasks in my programming language cheat sheets. ● LeetCode challenges. ● International Mathematical Olympiad (IMO) papers. Area where AIs haven't done well, even though we probably have the technology already, probably just oversights: ● Name examples of music with these properties. ● List words with these pronunciation properties. I.e. programmers would benefit from knowledge of the 44+1 phonemes of English, and their IPA representations. See Grapheme-Phoneme Correspondences - English.