London Futurists

The launch of GPT-4 on 14th March has provoked concerns and searching questions, and nowhere more so than in the education sector. Earlier this month, the share price of US edutech company Chegg halved when its CEO admitted that GPT technology was a threat to its business model.
Looking ahead, GPT models seem to put flesh on the bones of the idea that all students could one day have a personal tutor as effective as Aristotle, who was Alexander the Great’s personal tutor. When that happens, students should leave school and university far, far better educated than we did.
Donald Clark is the ideal person to discuss this with. He founded Epic Group in 1983, and made it the UK’s largest provider of bespoke online education services before selling it in 2005. He is now the CEO of an AI learning company called WildFire, and an investor in and Board member of several other education technology businesses. In 2020 he published a book called Artificial Intelligence for Learning.
Selected follow-ups:
https://donaldclarkplanb.blogspot.com/
https://www.ted.com/talks/sal_khan_how_ai_could_save_not_destroy_education
https://www.gatesnotes.com/The-Age-of-AI-Has-Begun
https://www.amazon.co.uk/Case-against-Education-System-Waste/dp/0691196451/
https://www.amazon.co.uk/Head-Hand-Heart-Intelligence-Over-Rewarded/dp/1982128461/
Topics addressed in this episode include:
*) "Education is a bit of a slow learner"
*) Why GPT-4 has unprecedented potential to transform education
*) The possibility of an online universal teacher
*) Traditional education sometimes fails to follow best pedagogical practice
*) Accelerating "time to competence" via personalised tuition
*) Calum's experience learning maths
*) How Khan Academy and DuoLingo are partnering with GPT-4
*) The significance of the large range of languages covered by ChatGPT
*) The recent essay on "The Age of AI" by Bill Gates
*) Students learning social skills from each other
*) An imbalanced societal focus on educating and valuing "head" rather than "heart" or "hand"
*) "The case against education" by Bryan Caplan
*) Evidence of wide usage of ChatGPT by students of all ages
*) Three gaps between GPT-4 and AGI, and how they are being bridged by including GPT-4 in "ensembles"
*) GPT-4 has a better theory of physics than GPT 3.5
*) Encouraging a generative AI to learn about a worldview via its own sensory input, rather than directly feeding a worldview into it
*) Pros and cons of "human exceptionalism"
*) How GPT-4 is upending our ideas on the relation between language and intelligence
*) Generative AI, the "C skills", and the set of jobs left for humans to do
*) Custer's last stand?
*) Three camps regarding progress toward AGI
*) Investors' reactions to Italy banning ChatGPT (subsequently reversed)
*) Different views on GDPR and European legislation
*) Further thoughts on implications of GPT-4 for the education industry
*) Shocking statistics on declining enrolment numbers in US universities
*) Beyond exclusivity: "A tutorial system for everybody"?
*) A boon for Senegal and other countries in the global south?
Music: Spike Protein, by Koi Discovery, available under CC0 1.0 Public Domain Declaration

One area of technology that is frequently in the news these days is rejuvenation biotechnology, namely the possibility of undoing key aspects of biological aging via a suite of medical interventions. What these interventions target isn't individual diseases, such as cancer, stroke, or heart disease, but rather the common aggravating factors that lie behind the increasing prevalence of these diseases as we become older.
Our guest in this episode is someone who has been at the forefront for over 20 years of a series of breakthrough initiatives in this field of rejuvenation biotechnology. He is Dr Aubrey de Grey, co-founder of the Methuselah Foundation, the SENS Research Foundation, and, most recently, the LEV Foundation - where 'LEV' stands for Longevity Escape Velocity.
Topics discussed include:
*) Different concepts of aging and damage repair;
*) Why the outlook for damage repair is significantly more tangible today than it was ten years ago;
*) The role of foundations in supporting projects which cannot receive funding from commercial ventures;
*) Questions of pace of development: cautious versus bold;
*) Changing timescales for the likely attainment of robust mouse rejuvenation ('RMR') and longevity escape velocity ('LEV');
*) The "Less Death" initiative;
*) "Anticipating anticipation" - preparing for likely sweeping changes in public attitude once understanding spreads about the forthcoming available of powerful rejuvenation treatments;
*) Various advocacy initiatives that Aubrey is supporting;
*) Ways in which listeners can help to accelerate the attainment of LEV.
Music: Spike Protein, by Koi Discovery, available under CC0 1.0 Public Domain Declaration
Some follow-up reading:
https://levf.org
https://lessdeath.org

This episode is a bit different from the usual, because we are interviewing Calum's boss. Calum says that mainly to tease him, because he thinks the word “boss” is a dirty word.
His name is Daniel Hulme, and this is his second appearance on the podcast. He was one of our earliest guests, long ago, in episode 8. Back then, Daniel had just sold his AI consultancy, Satalia, to the advertising and media giant WPP. Today, he is Chief AI Officer at WPP, but he is joining us to talk about his new venture, Conscium - which describes itself as "the world's first applied AI consciousness research organisation".
Conscium states that "our aim is to deepen our understanding of consciousness to pioneer efficient, intelligent, and safe AI that builds a better future for humanity".
Also joining us is Ted Lappas, who is head of technology at Conscium, and he is also one of our illustrious former guests on the podcast.
By way of full disclosure, Calum is CMO at Conscium, and David is on the Conscium advisory board.
Selected follow-ups:

• Conscium
• Satalia
• Six categories of application of AI
• Six singularities - TEDx talk by Daniel Hulme
• Professor Mark Solms
• Professor Karl Friston
• A recent paper on different theories of consciousness, by Patrick Butlin, Robert Long, et al
• Professor Nicola Clayton
• Professor Jonathan Birch
• WPP
• The Conscious AI meetup
• AI for organisations - Previous episode featuring Daniel Hulme
• How to use GPT-4 yourself - Previous episode featuring Ted Lappas

Music: Spike Protein, by Koi Discovery, available under CC0 1.0 Public Domain Declaration

The UK government has announced plans for a global AI Safety Summit, to be held in Bletchley Park in Buckinghamshire, outside London, on 1st and 2nd of November. That raises the importance of thinking more seriously about potential scenarios for the future of AI. In this episode, co-hosts Calum and David review Calum's concept of the Economic Singularity - a topic that deserves to be addressed at the Bletchley Park Summit.
Selected follow-ups:
https://www.gov.uk/government/news/uk-government-sets-out-ai-safety-summit-ambitions
https://calumchace.com/the-economic-singularity/
https://transpolitica.org/projects/surveys/anticipating-ai-30/
Topics addressed in this episode include:
*) The five themes announced for the AI Safety Summit
*) Three different phases in the future of AI, and the need for greater clarity about which risks and opportunities apply in each phase
*) Two misconceptions about the future of joblessness
*) Learning from how technology pushed horses out of employment
*) What the word 'singularity' means in the term "Economic Singularity"
*) Sources of meaning, beyond jobs and careers
*) Contrasting UBI and UGI (Universal Basic Income and Universal Generous Income)
*) Two different approaches to making UGI affordable
*) Three forces that are driving prices downward
*) Envisioning a possible dual economy
*) Anticipating "the great churn" - the accelerated rate of change of jobs
*) The biggest risk arising from technological unemployment
*) Flaws in the concept of GDP (Gross Domestic Product)
*) A contest between different narratives
*) Signs of good reactions by politicians
*) Recalling Christmas 1914
*) Suspension of "normal politics"
*) Have invitations been lost in the post?
*) 16 questions about what AI might be like in 2030
Music: Spike Protein, by Koi Discovery, available under CC0 1.0 Public Domain Declaration

How likely is it that, by 2030, someone will build artificial general intelligence (AGI)?
Ross Nordby is an AI researcher who has shortened his AGI timelines: he has changed his mind about when AGI might be expected to exist. He recently published an article on the LessWrong community discussion site, giving his argument in favour of shortening these timelines. He now identifies 2030 as the date by which it is 50% likely that AGI will exist. In this episode, we ask Ross questions about his argument, and consider some of the implications that arise.
Article by Ross: https://www.lesswrong.com/posts/K4urTDkBbtNuLivJx/why-i-think-strong-general-ai-is-coming-soon
Effective Altruism Long-Term Future Fund: https://funds.effectivealtruism.org/funds/far-future
MIRI (Machine Intelligence Research Institution): https://intelligence.org/
00.57 Ross’ background: real-time graphics, mostly in video games
02.10 Increased familiarity with AI made him reconsider his AGI timeline
02.37 He submitted a grant request to the Effective Altruism Long-Term Future Fund to move into AI safety work
03.50 What Ross was researching: can we make an AI intrinsically interpretable?
04.25 The AGI Ross is interested in is defined by capability, regardless of consciousness or sentience
04.55 An AI that is itself "goalless" might be put to uses with destructive side-effects
06.10 The leading AI research groups are still DeepMind and OpenAI
06.43 Other groups, like Anthropic, are more interested in alignment
07.22 If you can align an AI to any goal at all, that is progress: it indicates you have some control
08.00 Is this not all abstract and theoretical - a distraction from more pressing problems?
08.30 There are other serious problems, like pandemics and global warming, but we have to solve them all
08.45 Globally, only around 300 people are focused on AI alignment: not enough
10.05 AGI might well be less than three decades away
10.50 AlphaGo surprised the community, which was expecting Go to be winnable 10-15 years later
11.10 Then AlphaGo was surpassed by systems like AlphaZero and MuZero, which were actually simpler, and more flexible
11.20 AlphaTensor frames matrix multiplication as a game, and becomes superhuman at it
11.40 In 2018, the Transformer paper was published, but no-one forecast GPT-3’s capabilities
12.00 This year, Minerva (similar to GPT-3) got 50% correct on the math dataset: high school competition math problems
13.16 Illustrators now feel threatened by systems like Dall-E, Stable Diffusion, etc
13.30 The conclusion is that intelligence is easier to simulate than we thought
13.40 But these systems also do stupid things. They are brittle
18.00 But we could use transformers more intelligently
19.20 They turn out to be able to write code, and to explain jokes, and do maths reasoning
21:10 Google's Gopher AI
22.05 Machines don’t yet have internal models of the world, which we call common sense
24.00 But an early version of GPT-3 demonstrated the ability to model a human thought process alongside a machine’s
27.15 Ross’ current timeline is 50% probability of AGI by 2030, and 90+% by 2050
27:35 Counterarguments?
29.35 So what is to be done?
30.55 If convinced that AGI is coming soon, most lay people would probably demand that all AI research stops immediately. Which isn’t possible
31.40 Maybe publicity would be good in order to generate resources for AI alignment. And to avoid a backlash against secrecy
33.55 It would be great if more billionaires opened their wallets, but actually there are funds available for people who want to work on the problem
34.20 People who can help would not have to take a pay cut to work on AI alignment
Audio engineering by Alexander Chace
Music: Spike Protein, by Koi Discovery, available under CC0 1.0 Public Domain Declaration