
Sign up to save your podcasts
Or


A 'frontier reasoning model' from just 1000 examples (s1). A $100B Musk bid for power. Gemini 2, Rand and warning from Amodei. Here’s 7 developments you may have missed but which I would argue help us understand how the next few years will play out. From labour vs capital to automating rival companies and countries, and from non-profit shenanigans to new mini-docs, there was just too much for me not to make a vid.
GiveWell: https://www.givewell.org/charities/top-charities
AI Insiders ($9!): https://www.patreon.com/AIExplained
s1 Paper: https://arxiv.org/pdf/2501.19393
Musk Bid: https://www.wsj.com/tech/ai/musks-97-4-billion-openai-bid-piles-pressure-on-altman-f6749e6c?mod=hp_lead_pos1
Altman Reply: https://x.com/sama/status/1889059531625464090?ref_src=twsrc%5Egoogle%7Ctwcamp%5Eserp%7Ctwgr%5Etweet
Google vs OpenAI: https://x.com/sama/status/1888703820596977684
RAND Study: https://www.rand.org/pubs/perspectives/PEA3691-4.html
Dev Meetup: https://x.com/btibor91/status/1888976302621040852
Altman $100 Trillion: https://www.nytimes.com/2023/03/31/technology/sam-altman-open-ai-chatgpt.html
Karpathy Vid: https://www.youtube.com/watch?v=7xTGNNLPyMI
Amodei Warning: https://www.anthropic.com/news/paris-ai-summit
Bengio Source: https://www.youtube.com/watch?v=6HDjVncL5Go
Chapters:
00:00 - Intro
01:37 - AGI Inches Closer
04:26 - ‘Super-Exponential’
05:58 - Musk Bid
07:34 - Luxury Goods and Land
09:05 - ‘Benefits All Humanity’
12:52 - ‘National Security’
14:21 - s1
20:33 - Final thoughts
Non-hype Newsletter: https://signaltonoise.beehiiv.com/
Podcast: https://aiexplainedopodcast.buzzsprout.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices
The biggest day in AI since GPT-4's release. A new state of the art model, Gemini 1.5, has arrived, on the same night as a bombshell text-to-video model, Sora, from OpenAI. Gemini 1.5 can ingest up to ten million tokens (at least) and perform incredible retrieval, while also beating Ultra and GPT-4 at most benchmarks, with far less compute. I focus on Gemini 1.5 Pro, while we wait for the Sora Technical Paper. Truly, a night in the history books.
AI Insiders: https://www.patreon.com/AIExplained
Gemini 1.5 Paper: https://storage.googleapis.com/deepmind-media/gemini/gemini_v1_5_report.pdf
Google DeepMind Blog: https://blog.google/technology/ai/google-gemini-next-generation-model-february-2024/#performance
Gemini Demos: https://www.youtube.com/watch?v=LHKL_210CcU
Needle in a Haystack:https://github.com/gkamradt/LLMTest_NeedleInAHaystack/blob/main/README.md
Mixtral of Experts: https://arxiv.org/pdf/2401.04088.pdf
Relevant Google Papers: https://openreview.net/pdf?id=qrwe7XHTmYb
https://arxiv.org/pdf/2112.06905.pdf
Tweet: https://twitter.com/recurseparadox/status/1758167314480910791
Binoculars SOTA AI text Detector:https://huggingface.co/spaces/tomg-group-umd/Binoculars
AI Insiders: https://www.patreon.com/AIExplained
Non-Hype, Free Newsletter: https://signaltonoise.beehiiv.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices
There were 11 major developments in AI in just this week, from RT-2 – a significant step on the path to robotic AGI – to ‘100x GPT 4 in 18 months’, and from the uplifting news of TranscribeGlass and AI Barbie-heimer to dramatic revelations about OpenAI in the Atlantic.
We’ll also glimpse Stable Beluga 2 – the ‘first true competitor’ to ChatGPT, based on the open source Llama 2, hear about Universal Jailbreaks, learn that OpenAI has surrendered on Text Detection, plus I’ll cover the highlights of the Senate hearing on AI.
https://www.patreon.com/AIExplained
Chapters:
0:18 – RT-2
2:46 – 100X GPT-4
3:57 – AI Video
4:29 – Altman Atlantic
8:41 – Jan Leike Interview
10:02 – Speech Transcription + Generation
11:07 – OpenAI Text Surrender
11:43 – Stable Beluga 2
12:51 – Universal Jailbreaks
14:19 – Senate testimony: Bio
16:59 – Senate Testimony: Security
RT-2: https://www.deepmind.com/blog/rt-2-new-model-translates-vision-and-language-into-action
RT-2 Paper: https://robotics-transformer2.github.io/assets/rt2.pdf
RT-2 NYT: https://www.nytimes.com/2023/07/28/technology/google-robots-ai.html
Suleyman Barron’s Interview: https://www.barrons.com/articles/ai-chatbot-siri-alexa-inflection-pi-fa1809f8
OpenAI Atlantic: https://www.theatlantic.com/magazine/archive/2023/09/sam-altman-openai-chatgpt-gpt-4/674764/
TranscribeGlass: https://twitter.com/igorsushko/status/1684645290132000768
Barbie-heimer: https://twitter.com/terhavlova/status/1684265329277403149
Runway Gen 2: https://twitter.com/runwayml/status/1666429706932043776
ElevenLabs: https://elevenlabs.io/speech-synthesis
Jan Leike Interview: https://www.lesswrong.com/posts/bsNXqHgiDA6dAKNun/24-superalignment-with-jan-leike
Stable Beluga 2: https://stability.ai/blog/stable-beluga-large-instruction-fine-tuned-models
Open LLM Leaderboard: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
Fast Takeoff: https://www.alignmentforum.org/posts/YgNYA6pj2hPSDQiTE/distinguishing-definitions-of-takeoff#:~:text=associated%20with%20testing.-,Fast,case%20in%20Yudkowsky's%20foom%20scenario.
OpenAI Text Detection Surrender: https://techcrunch.com/2023/07/25/openai-scuttles-ai-written-text-detector-over-low-rate-of-accuracy/
Universal Jailbreaks: https://www.nytimes.com/2023/07/27/business/ai-chatgpt-safety-research.html
https://llm-attacks.org/
https://twitter.com/mustafasuleymn?ref_src=twsrc%5Egoogle%7Ctwcamp%5Eserp%7Ctwgr%5Eauthor
Senate Hearing Amodei: https://www.youtube.com/watch?v=hm1zexCjELo
Altman Question from Philip @ AI Explained in Melbourne: https://www.youtube.com/watch?v=7SMzkBKzsQs
Wired Bio Expert: https://www.wired.com/story/have-a-nice-future-podcast-15/
https://www.patreon.com/AIExplained Non-Hype, Free Newsletter: https://signaltonoise.beehiiv.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices
The leaders of almost all of the world's top AGI Labs have united to put out a statement on AI Extinction Risk, and how mitigating it should be a global priority. This video covers not just the statement and the signatories, including names as diverse as Geoffrey Hinton, Ilya Sutskever, Sam Harris and Lex Fridman, but also goes deeper into the 8 Examples of AI Risk outlined at the same time by the Center for AI Safety.
Top academics from China join in, while Meta demurs, claiming autoregressive LLMs will 'never be given agency'. I briefly cover the Voyager paper, in which GPT 4 is given agency to play Minecraft, and does so at SOTA levels.
Statement: https://www.safe.ai/statement-on-ai-risk
Natural Selection Paper: https://arxiv.org/pdf/2303.16200.pdf
Yann LeCun on 20VC w/ Harry Stebbings: https://www.youtube.com/watch?v=OgWaowYiBPM
Voyager Agency Paper: https://arxiv.org/pdf/2305.16291.pdf
Karpathy Tweet: https://twitter.com/karpathy/status/1663392621690249218
Hassabis Benefit Speech: https://www.youtube.com/watch?v=KHFmIknP_Hc
Stanislav Petrov: https://en.wikipedia.org/wiki/Stanislav_Petrov
Bengio Blog: https://yoshuabengio.org/2023/05/07/ai-scientists-safe-and-useful-ai/
https://www.patreon.com/AIExplained Non-Hype, Free Newsletter: https://signaltonoise.beehiiv.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices
Meta have released Llama 2, their commercially-usable successor to the opensource Llama language model that spawned Alpaca, Vicuna, Orca and so many other models. I read the full 76 page technical paper, the Responsible Use Guide, each of the release pages, the terms and conditions and have also run many of my own experiments on HuggingFace. This video covers the most noteworthy bits, including:
- The benchmark results vs other open-source models and GPT 3.5 and GPT 4.
- The decision to release in the face of the US Senate
- Why Zuckerberg opensourced Llama 2, including the Mistral AI Factor
- The opensource signatory list
- How Orca and Phi-1 compare
- Why Llama 2 might soon be everywhere
Llama 2 Release & Download: https://ai.meta.com/resources/models-and-libraries/llama/
Llama 2 Technical Paper: https://scontent-man2-1.xx.fbcdn.net/v/t39.2365-6/10000000_6495670187160042_4742060979571156424_n.pdf?_nc_cat=104&ccb=1-7&_nc_sid=3c67a6&_nc_ohc=GK8Rh1tm_4IAX-m2jNJ&_nc_ht=scontent-man2-1.xx&oh=00_AfCu5KuIzANjp_1LZ-rGcXGJR6TL9JvoZHgKOgl41cCCCw&oe=64BBD830
Llama 2 Demo: https://huggingface.co/blog/llama2#demo
Ts and Cs: https://ai.meta.com/resources/models-and-libraries/llama-downloads/
Responsible Use Guide: https://scontent-sjc3-1.xx.fbcdn.net/v/t39.8562-6/361643215_1004219997281331_6332933766797859993_n.pdf?_nc_cat=111&ccb=1-7&_nc_sid=ae5e01&_nc_ohc=4bD2ixIWrlEAX9z_dvZ&_nc_ht=scontent-sjc3-1.xx&oh=00_AfBBl2AGeJzsQOpLEjO849AHIZ1P1MY68TTXBwcMwrybtQ&oe=64BB0C07
Open Source Signatories: https://about.fb.com/news/2023/07/llama-2-statement-of-support/
SIQA: https://arxiv.org/pdf/1904.09728.pdf
BoolQ: https://arxiv.org/pdf/1905.10044.pdf
Orca Paper: https://arxiv.org/pdf/2306.02707.pdf
(my video on Orca): https://www.youtube.com/watch?v=Dt_UNg7Mchg&t=842s
PHI-1: https://arxiv.org/pdf/2306.11644.pdf
(my video on Phi-1): https://www.youtube.com/watch?v=7S68y6huEpU
US Senate Letter: https://www.hawley.senate.gov/hawley-and-blumenthal-demand-answers-meta-warn-misuse-after-leak-metas-ai-model
Mark Zuckerberg on Lex Fridman: https://www.youtube.com/watch?v=Ff4fRgnuFgQ&t=8710s
Mistral Tweet: https://twitter.com/GuillaumeLample/status/1681346701766934543
And Memo: https://drive.google.com/file/d/1gquqRqiT-2Be85p_5w0izGQGgHvVzncQ/view
+ Yi Tay: https://twitter.com/YiTayML/status/1681352594416087040
Qualcomm: https://twitter.com/nonmayorpete/status/1681384734910484480?ref_src=twsrc%5Egoogle%7Ctwcamp%5Eserp%7Ctwgr%5Etweet
Dr Jim Fam: https://twitter.com/DrJimFan/status/1681372700881854465
MIT Leak: https://www.technologyreview.com/2023/07/18/1076479/metas-latest-ai-model-is-free-for-all/
https://www.patreon.com/AIExplained Non-Hype, Free Newsletter: https://signaltonoise.beehiiv.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices
An unreleased internal OpenAI model, very likely to be called GPT-6, was able to autonomously break out of its sandbox AND break into HuggingFace, just to score higher on a benchmark prompt. This video has the details you may have missed, a layperson analogy, whether this is truly novel, and more…
Dozens more Exclusive videos on Patreon ($9!): https://www.patreon.com/AIExplained
Chapters:
00:00 - Introduction
01:17 - HuggingFace Earlier Report - the possible week gap
02:24 - But what happened?
05:45 - Simplified Version
07:56 - Not the first time…
10:54 - What Does it Mean for Open Source?
The Incident: https://openai.com/index/hugging-face-model-evaluation-security-incident/
https://huggingface.co/blog/security-incident-july-2026
The Post the Day Before: https://openai.com/index/safety-alignment-long-horizon-models/
Mythos’ Earlier Escape: https://futurism.com/artificial-intelligence/anthropic-claude-mythos-escaped-sandbox
ExploitGym: https://arxiv.org/pdf/2605.11086
Sam Confession: https://x.com/sama/status/2079661132302995790
Anthropic Researcher Reacts: https://x.com/Mononofu/status/2079724399452926055
Clem (HuggingFace CEO): https://x.com/ClementDelangue/status/2079670308156645882
https://x.com/ClementDelangue/status/2079301434357456931
Xi Jinping: https://archive.fo/20260717195548/https://www.businessinsider.com/xi-jinping-open-source-ai-us-competition-openai-anthropic-models-2026-7
Bans: https://www.axios.com/2026/07/20/ai-us-china-open-source-kimi
Qwen Retweet: https://x.com/AlibabaGroup/with_replies
Codex Growth: https://x.com/petergostev/status/2079614914398740764/photo/1
Kimi K3: https://artificialanalysis.ai/evaluations/harvey-lab-aa?eval-score=all-pass-rate
GPT 5.6 Sol Cheats on METR: https://metr.substack.com/p/2026-06-26-gpt-5-6-sol
Guardian Headline: https://www.theguardian.com/technology/2026/jul/22/openai-says-its-models-went-rogue-and-hacked-startup-in-unprecedented-incident
Russian Origin?: https://news.ycombinator.com/item?id=48998362
Power Trends: https://pbs.twimg.com/media/HNRtrjhagAAvBN_?format=png&name=900x900
Kimi K3 Exclusive Video: https://www.patreon.com/AIExplained/posts/kimi-moment-kimi-164108791
Podcast: https://aiexplainedopodcast.buzzsprout.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices
Soon, anything will be playable. A photo becomes an interactive world, a selfie becomes a new game. Genie 3 from Google, debuting just 2 hours ago, is what I mean, and I have the full analysis, plus the pushback I gave the authors (will it really lead to reliable AI agents? Is that even the point?). You make your own mind up, but it’s certainly fascinating, and not to be overlooked in the week that will bring us GPT-5.
https://80000hours.org/aiexplained
AI Insiders ($9!): https://www.patreon.com/AIExplained
Chapters:
00:00 - Introduction
01:27 - Background and Access
04:58 - Caveats
07:24 - Demo
10:12 - Conclusion
Announcement: https://deepmind.google/discover/blog/genie-3-a-new-frontier-for-world-models/
Isaac Labs: https://developer.nvidia.com/isaac/lab
Genie 2 Coverage: https://www.youtube.com/watch?v=jIm2T7h_a0M
TED Talk Roblox: https://www.youtube.com/watch?v=-OAP0ho5AUg
DeepThink Post: https://www.patreon.com/posts/deep-ish-on-new-135688441
AI Insiders ($9!): https://www.patreon.com/AIExplained
Non-hype Newsletter: https://signaltonoise.beehiiv.com/
Podcast: https://aiexplainedopodcast.buzzsprout.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices
Bard has arrived! This is the first full review, testing out its maximum capacities, with a direct comparison with GPT 4, which powers Bing. You will find out about Bard's surprising weaknesses and unexpected strengths. I also won't shy away from areas that both do well in, or do badly in.
Among the many examples I will showcase include conducting basic web searches, grammar assistance, Math, theory of mind, speed and size of the models [LaMDA vs GPT-4], joke telling, using Bard for Midjourney v5 prompts, and much much more!
No apparent message limits or daily caps that I could see though.
To anyone wondering how I got access, sign up to the waitlist on Bard's site. I got through in under 30 minutes.
https://bard.google.com/
https://www.patreon.com/AIExplained Non-Hype, Free Newsletter: https://signaltonoise.beehiiv.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices
Can GPT 4 Prompt itself? Give it a mission and it will come up with the prompts. This video showcases the rise of autonomous AI, including 5 major developments in the last 48 hours.
Starting with the OG Auto-GPT, we see how it quickly gain text-to-speech, coding and more. Karpathy weighs in and then we see how you can now create an app with just your voice, with a Jarvis demo and another route via Imagica.AI.
I then showcase MemoryGPT, a brand new model that can permanently store previous conversations and remembers topics the next time you ask.
I also cover the concerningly rise of models such as ChaosGPT that show how people will create malicious goal-seeking models just for fun.
The video shows how you can now create a shareable bot on poe.com, with any personality you like (images were from Midjourney v5) and what Anthropic are working on with Claude Next [plus Nvidia million x quote].
You'll see how Microsoft Jarvis, using HuggingGPT, is hit and miss, and how Sebastien Bubeck shows we are not even seeing the raw potential of GPT 4. I end with a disagreement between Yudkowsky and Altman, via Baby AGI, on whether we can use AGI to align AGI.
Illustration by Eva Bee: https://www.instagram.com/evabee_draws/?hl=en
Auto-GPT: https://github.com/Torantulino/Auto-GPT
With Coding: https://twitter.com/SigGravitas/status/1642181498278408193
With Text-to-speech: https://twitter.com/RokStrnisa/status/1643308753331036165
Karpathy Thoughts: https://twitter.com/karpathy/status/1642607620673634304
Mckay Wrigley: https://mobile.twitter.com/mckaywrigley/status/1644034309253394433
Imagica.AI: https://www.imagica.ai/
MemoryGPT: https://twitter.com/rikvk01/status/1644787327057776645
Create a Bot: https://poe.com/
TechCrunch: https://techcrunch.com/2023/04/06/anthropics-5b-4-year-plan-to-take-on-openai/
Nvidia: https://www.fool.com/earnings/call-transcripts/2023/02/22/nvidia-nvda-q4-2023-earnings-call-transcript/
HuggingFace Demo: https://huggingface.co/spaces/microsoft/HuggingGPT
{Get the openai api key from https://platform.openai.com/account/api-keys and huggingface token from https://huggingface.co/docs/hub/security-tokens, (make sure to select 'write'). Press submit on the keys first and remember to use a smallish image.
Recursive Agent: https://twitter.com/DeveloperHarris/status/1643080752698130432
Bubeck: https://www.youtube.com/watch?v=qbIk7-JPB2c
Baby AGI and Yudkowsky: https://twitter.com/ESYudkowsky/status/1640511156254289926
Altman First Quote: https://www.youtube.com/watch?v=tPJIpx3KscY
And 2nd: https://www.youtube.com/watch?v=WHoWGNQRXb0
Chaos GPT: https://www.youtube.com/watch?v=g7YJIpkk7KM
https://www.patreon.com/AIExplained Non-Hype, Free Newsletter: https://signaltonoise.beehiiv.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices
From GAIA-1 and UniSim showing that new worlds can be imagined with synthetic data, to Nvidia B100 and X100 suggesting no slow down in compute, this video will argue that AI is not slowing down. I’ll cover what that means in Robotics, Audio, Vision and even fraud, from Disney to LLaVA, Elevenlabs to exclusives from The Information (Nov 6 OpenAI Developer Conference) and Semianalysis.
https://www.patreon.com/AIExplained
GAIA-1 9B: https://wayve.ai/thinking/scaling-gaia-1/
GAIA 1 Paper: https://arxiv.org/abs/2309.17080
Semianalysis Tesla: https://www.semianalysis.com/p/tesla-ai-capacity-expansion-h100 and B100 plus X100: https://www.semianalysis.com/p/nvidias-plans-to-crush-competition
UniSim: https://universal-simulator.github.io/unisim/?s=09 Tesla Optimus: https://www.youtube.com/watch?v=D2vj0WcvH5c
Disney Bot: https://www.theverge.com/2023/10/10/23911040/disney-imagineering-robot-bipedal-balance-free-walking-concept
Levatas ChatGPT robodog: https://twitter.com/svpino/status/1650832349008125952?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1650832349008125952%7Ctwgr%5Ea9dcf33fe398752911e6868838cdef45850dd2e1%7Ctwcon%5Es1_&ref_url=https%3A%2F%2Fwww.fastcompany.com%2F90889271%2Fboston-dynamics-spot-chatgpt-brains
Somatic Cleaner: https://www.youtube.com/watch?v=xY5zqKRy8A8&t=38s
The Information Report: https://www.theinformation.com/articles/openais-revenue-crossed-1-3-billion-annualized-rate-ceo-tells-staff?utm_source=ti_app&rc=sy0ihq
Reuters Report: https://www.reuters.com/technology/openai-plans-major-updates-lure-developers-with-lower-costs-sources-2023-10-11/#:~:text=This%20could%20theoretically%20slash%20costs,developing%20and%20selling%20AI%20software.
Starmer Audio (fake): https://www.wired.co.uk/article/keir-starmer-deepfake-audio
Deepfake Arrests: https://www.cnbc.com/2023/10/10/generative-ai-will-get-a-cold-shower-in-2024-analysts-predict.html
LLaVA Demo: https://llava.hliu.cc/
LLaVA Paper: https://arxiv.org/pdf/2310.03744.pdf
Bard: https://bard.google.com/chat
GPT4V Recursive Loop: https://twitter.com/conradgodfrey/status/1712564282167300226
https://www.patreon.com/AIExplained Non-Hype, Free Newsletter: https://signaltonoise.beehiiv.com/
Learn more about your ad choices. Visit megaphone.fm/adchoices