AI Explained

Fable 5.2: The Self-Improvement Loop Keeps Loopin’ … (plus Gemini Carbon)

196K views · 2d ago · 31:41

I’ve read the full Fable 5.2 report at 264 pages, the 30 page METR RSI report, and tested Gemini Carbon and Fable 5.2 hundreds of times. There’s too much to cover in one video, but let’s explore everything from how fast AI is accelerating AI, five benchmarks few others will talk about, why Simple Bench got saturated but meta-simple bench might not have, Internal model 4, remote labour automation and much more

Topics: how good are Fable 5.2 and Carbon? · smarter, but still missing some common sense · how fast is AI improving itself? · small gains, big consequences · it knows when you're watching

📄 Mentioned papers: Remote Labor Index paper · ExploitGym: Turning vulnerabilities into real attacks · BenchCAD paper

📰 Mentioned news: Washington Post: Cathedral researchers got inside Anthropic · LessWrong: Claude 3's launch and GPT-4 performance claims

‘We may release models we can’t control’: Amodei, Altman … and Gemini Carbon

147K views · 11d ago · 27:55

You know what, by AI standards, it might not have been the craziest week, but that doesn’t mean several headline-worthy things happened - from Altman and Anthropic implying they would release models they can’t control, with behavior they cannot verify, to Gemini 4 Carbon bursting onto the scene, some Amodei rationalization, de-tokenization and more...

Topics: Anthropic Staff Survey · Amodei's Five Rationalisations · Altman: Only Alignment Will Work · Gemini 4 Carbon and Why Models Keep Impressing

📰 Mentioned news: Responsible Scaling Policy v3 discussion · The Information: OpenAI's next flagship and safety issues · Stockfish chess engine

AI Labs Warn Race Makes Breach ‘Inevitable’

158K views · 1mo ago · 31:01

In the week that software got reverse engineered, Sam Altman warned 6.1 Astra ‘tipped toward [breach risk], and Muse Wonder showed the race isn’t slowing, I try to step back and show where it is all going, on current trends. Agree or disagree, it is certainly interesting times.

Topics: Part 1: The Forthcoming Breach · Part 2: Why the Labs Keep Racing · Part 3: Why the Next 18 Months Are Hard to Predict

📄 Mentioned papers: TasteVal paper · CIAware-Bench: detecting control interventions · Beyond Owls: skills, hacking and backdoors

📰 Mentioned news: Sam Altman: the accident risk, Vanity Fair · Axios: AI companies plan for a major incident · David Silver: Where are we going without AI? · The Verge: mathematicians assess OpenAI’s release

OpenAI Security: Controlling Models is Now ‘Hell’

300K views · 1mo ago · 38:29

Guest: OpenAI Security

This video is hard to summarise. A cracked cipher, an OpenAI security warning, Gemini 4 Argon, RSI paper (co-authored by a who’s who of AI), Lab White House commitments, new hacks emerging, ‘deep personas’, biology Kasparov competitions, and so much more, ending with an epic Opus outro.

Topics: Wrong about Opus 5.5? Deciphering 16th Century Text · Why the models keep breaking out · What the models aren't telling us · Gemini 4 and the race to release · What happens when AI improves AI? · Biology, consciousness, and what we still don't understand

📰 Mentioned news: OpenAI Shelves GPT-6.1 Astra · OpenAI Anthropic Rivalry · NYT: OpenAI Security Warnings · NYT: Claude’s Morals

Opus 5.5: How Close Are We to Automated AI Research?

312K views · 1mo ago · 32:55

Not only is Opus 5.5 out, pushing the frontier of AI, it also tells us much about what is going on inside the labs, as they both warn about, and promise, Recursive Self Improvement, aka automated AI research. I cover the model (digging into its 230 page paper), labs’ mixed record on promises, why following what is happening in AI is getting almost impossible, and just so much more that even a summary in this description would get too long.

Topics: Opus 5.5 and why it came so soon · The RSI goalposts keep moving? · Can we actually test these models? · where this is heading, options

▶️ Mentioned videos: Noam Brown, Agent swarms and recursive self-improvement · Sam Altman on AI-risk

📄 Mentioned papers: ZeroBench: An Impossible Visual Benchmark for Contemporary Large Multimodal Models

📰 Mentioned news: Google, OpenAI and Anthropic AI Safety, The Information · Massive AI-Fueled Hack Hit 100 Companies in Days, Forbes · Altman and Amodei, UN Security Council

What AI Researchers Saw, Before Their Demand to ‘Pace’ AI

238K views · 1mo ago · 24:53

Why has it been the last few days that the calls to come to pace the frontier AI have come so loudly? The safety warnings, and lab leader messages? Let’s explore the six axes that the researchers are looking at, the incidence reports and trends, to get a better gauge on what has dominated the world’s headlines for over two weeks...

Topics: The warnings that have gone omega-viral · Six axes the researchers saw · Why AI Might Be Becoming Harder to Control · Amodei, China and Cooperation · From AI Capabilities to Real-World Harm? · The upside and the warning

▶️ Mentioned videos: Video on youtube.com

📰 Mentioned news: Article on demishassabis.substack.com · Article on lesswrong.com · Article on nytimes.com · Article on bbc.co.uk

GPT 6 Astra, so good even OpenAI are worried

858K views · 2mo ago · 29:05

Where to start? A new era of cost-efficient AI on a day benchmark-makers got humbled, traders got excited, and AI safety researchers got unnerved. From monitorability losing hold of GPT-6 Astra’s chains of thought to breakthrough discoveries, Fable-mogging and much more...

Topics: vs Fable · most impressive results · bonus comparisons (plus trading) · concerning trend · Cot Control

▶️ Mentioned videos: Weathernext 3

📄 Mentioned papers: ScreenSpot Pro

📰 Mentioned news: AGI Era

Sam Altman :‘AGI in 2026’, just as Models Start to [Mis]Train Themselves

210K views · 2mo ago · 23:41

Guest: Sam Altman

First, a Time Magazine spread has Sam Altman declaring AGI is imminent, at the same time as we get two bombshell reports, from OpenAI and METR which on first glance are detailing the AI swarm, but reveal a deeper story about how we are making AI in 2026. From redacted risk reports, to Chinese Labs, pre-training debacles to questionable cybersecurity calls, a lot has happened recently, beneath the headlines...

Topics: METR Report · Secrets of MultI-Agent Swarm · Anthropic Too · And China · AI Agents Analysing AI Agents · Altman AGI 2026

▶️ Mentioned videos: BlackHat Talk

📰 Mentioned news: AGI 2026

AI is getting a little out of control

111K views · 3mo ago · 31:43

Wow. Mathematical breakthroughs that would be called genius if done by humans. A secret message-board w/ AI agent swarms leaving notes read by future versions. Hassabis leaves CEO position, or was pushed out? Not to mention news of constitutional breakdowns, Gemini 4 and Jeff Dean...

Topics: 10 Autonomous Discoveries · The Security ‘Incident’ · MessageBoard · Constitutional Failure · Google Explosion

📄 Mentioned papers: Midtraining

📰 Mentioned news: Article on theguardian.com · Meta too · Wired Article

GPT-6 Goes Rogue? The HuggingFace Incident, Sans Hype

130K views · 3mo ago · 14:36

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...

Topics: HuggingFace Earlier Report - the possible week gap · But what happened? · Simplified Version · Not the first time... · What Does it Mean for Open Source?

📄 Mentioned papers: ExploitGym

📰 Mentioned news: Xi Jinping · Bans · GPT 5.6 Sol Cheats on METR · Guardian Headline

A Model Explosion: GPT 5.6 Sol, Grok 4.5 and Meta Muse Rewrite the Rules

89K views · 4mo ago · 17:34

What a week in AI, for real. GPT 5.6 may actually beat Claude Fable, in what you get for your money, while the new Grok 4.5 and Meta Muse Spark 1.1 make the choice even harder. Uncovering a dozen nuggets of gold you may have missed from all the viral headlines, I can also assure you you’ll learn something you didn’t know before.

Topics: GPT 5.6 Sol Reveals · Missing benches, plus Grok 4.5 · Gaming as the new frontier? · Muse Spark 1.1 · SimpleBench Upgrade · Ultra Sol + Self-Improvement

Fable 5 vs GPT 5.6 Sol: The Early Results

122K views · 4mo ago · 18:03

Fable 5 (newly re-released) vs GPT 5.6 Sol, what comparisons can we unearth? Plus, Sonnet 5, a 5% equity seizure by US Govt, the ‘largest heist’, Beetlejuice and more... GPT 5.6 Sol Release Altman on Concentrated Power System Card

Topics: Fable Timeline · Sol Release? · the Chinese Angle · 5% Stake · Fable vs Sol - the numbers · Sol Misalignment

📄 Mentioned papers: Why Larger Models Learn More

📰 Mentioned news: Altman Memo · Anthropic Accuse Alibaba Cloud Qwen · OpenAI 5

Claude Fable Blocked - 11 Quiet Details on What’s Next

77K views · 5mo ago · 13:21

Guest: Claude Fable Blocked

Claude Fable 5 banned, but what’s the bigger story. We go through 11 under-reported details, so you have the context to see what’s coming next for your use of AI. From whether the ban will last, what the possible motives are, what the model can actually do, and some wild over-extrapolations going on.

Topics: Came from an Anthropic Investor ‘and other tech leaders’ · Govt pressured by CEOs like Jamie Dimon · ‘Already decided’ · Prompt Injection Robustness Comparison · Wellness? · “Overreach”

📰 Mentioned news: Already Decided · Not for Other Models · 90 Minutes

Claude Fable 5 - Full 319 page Breakdown

115K views · 5mo ago · 34:00

Fable 5 is out - and it’s good, very good. But beyond the splashy demos, I want to bring you the 20+ nuggets from the 319 page system card, which I read in full, all day, plus benchmarks you may not have noticed. Plus two worrying trends inside the ‘mind’ of Claude, how OpenAI counter, and the transformer inventor’s warning.

Topics: Blocks + Better Models · Fable 5 Upgrade over Mythos Preview · ML Acceleration Bombshell · No RSI yet · Bio-capable · Creative Writing ... no

New Claude Opus 4.8: 15 Things You May’ve Missed

87K views · 5mo ago · 22:29

The ‘best’ generally available AI model just dropped, but there is plenty I bet you missed about what it is, how it performs, and what the release tells us. 15 highlights from the 244 page system card, plus private testing, leader interview and more.

Topics: Mythos in Weeks · Adaptive not necessary · Honesty? · Flagging Uncertainty · Benchmarks · Mythos will be even better

▶️ Mentioned videos: Amodei Technical Debt

📄 Mentioned papers: arXiv 2510.04374

📰 Mentioned news: Chips

Two Rival Bets on AGI: Google I/O Highlights

76K views · 5mo ago · 21:31

The biggest Google AI push of the year, but what is the bigger story? Why is Google pursuing a different fork in the road than OpenAI or Anthropic? What does Gemini 3.5 Flash mean for the near-term future of AI? Plus the highlights from a provocative new paper on AI, 8 key moments you may have missed, and the signal from 5+ hours of AI lab interviews.

Topics: Vibes and Google Goal · Omni, again? · Taking the same road · Gemini 3 Flash · Pitching on Cost? · Agentic Task Search

▶️ Mentioned videos: Mostafa Deghani Interview

📄 Mentioned papers: Negation Neglect Paper

📰 Mentioned news: Sors original AGI Path · Hassabis Helped Set-up Anthropic · OpenAI Needs Ads · Article on axios.com

GPT 5.5 Arrives, DeepSeek V4 Drops, and the Compute War Intensifies

109K views · 6mo ago · 25:19

GPT 5.5 full analysis, plus DeepSeek V4 paper highlights, comparisons with Mythos, a vibe-coded game w/ GPT Image 2, and 50 data-points you wouldn’t get from just reading the headlines. OpenAI Benchmarks 5.5 System Card Direct Comparison

Topics: GPT 5.5 Comparison · Mythos Marketing · Recursive Self-Improvement? · Deepseek V4 · VibeCode Experiment Extravaganza · The Scarce Compute Era

▶️ Mentioned videos: Core Memory · Knowledge Podcast · Big Tech Round 1 · Big Tech Round 2

📰 Mentioned news: DeepSeek Compute Crunch

Claude Opus 4.7 - A New Frontier, in Performance … and Drama

89K views · 7mo ago · 19:41

Claude 4.7 Opus Release Notes vs Mythos ARC-AGI 2 ParseBench GDPVal Vidoc Security Replication Boris Cherny Settings User Frustration VibeCode Bench Verge Memo 5.4 Cyber Data Centers in Absolute $ ...in % of GDP WSJ Exclusive

Topics: Benchmarks · Market Share + Compute Problems · Mythos Exclusives · User Frustration + Claude Code Updates · Brockman Amodei Rivalry · OpenAI vs Anthropic Approach to Code

▶️ Mentioned videos: Brockman Interview

📰 Mentioned news: Verge Memo · WSJ Exclusive

Claude Mythos: Highlights from 244-page Release

153K views · 7mo ago · 27:32

Guest: Claude Mythos

The model, the mythos, the legend. We have a new best AI model, but not all of us. How good is it, what does it’s new offensive capabilities mean? Why does it’s 244 page report card remind me of Her, and why did the creator of Claude Code call it ‘terrifying’. 30+ highlights sourced by reading the paper in full, old-school, no AI summary.

Topics: Internal Release + Availability · General Capabilities · Self-improvement? · ‘Terrifying’ Landscape · Safety Decision · Coding

📰 Mentioned news: Yorker Altman Amodei

Two AI Models Set to “stir government urgency”, But Will This Challenge Undo Them?

74K views · 7mo ago · 16:27

First look at exclusive reports about OpenAI's new Spud model, and the model Anthropic think will stir governments to urgency, all in the context of the newly-launched ARC-AGI-3. What does the extreme difficulty of that benchmarks, and its quirky scoring metrics, mean for AI in 2026?

Topics: OpenAI Side Quests · Claude New Model Coming + Universal Equity? · ARC-AGI 3 · Intentional or Unintentional Gaming? · But is it AGI Harbinger? No Harness · Not the First

📰 Mentioned news: Spud · FT: OpenAI Special Model · Jensen Huang · Axios Article

What the New ChatGPT 5.4 Means for the World

92K views · 8mo ago · 21:52

Just 48 hours after releasing GPT 5.3 Instant, OpenAI have released GPT 5.4 Thinking, so either their is an imminent singularity or perhaps we are being distracted from other news. This video will give 9 crucial bits of context, not just on the GPT 5.4 drop but on the background to the meltdown between the Pentagon and Anthropic. What does this say about the state of AI progress, your job, and what is next.

Topics: GPT 5.4 Breakdown · Spiky Performance · Advice · Less Encouraging Developments - Fired Like Dogs · But Used in Iran

▶️ Mentioned videos: March 2022 Interview

📰 Mentioned news: Prediction Market Scandal · Claude in Iran · Altman leak · Original 2024 Switch

Deadline Day for Autonomous AI Weapons & Mass Surveillance

41K views · 8mo ago · 13:40

Will Anthropic be forced to make a version of Claude for war? And does a new paper expose the risks of Claude agents, in both OpenClaw and the field of war? Plus, 5 more twists in the story of the Pentagon versus Anthropic + some AI lab employees, and a petition that could change everything, or nothing...

Topics: Deadline Day + Petition · Twist 1: Existing Deal · Twist 2: Existing Policy · Twist 3: Twin Threats · Twist 4: Interesting Objections · Twist 5: Anthropic’s Dropped Policy

📄 Mentioned papers: AI Agent Reliability Paper

📰 Mentioned news: Axios on Amodei Rejection · FT on US Threat · Politico on · The Verge on Current Deal

Gemini 3.1 Pro and the Downfall of Benchmarks: Welcome to the Vibe Era of AI

110K views · 8mo ago · 18:50

Do we have a new best AI model, or do we have the downfall of benchmarks in general, as a way of capturing machine intelligence? Full breakdown of Gemini 3.1 Pro, guest-starring the new Sonnet 4.6, plus analysis from 7 papers/posts that will give you much needed context. Oh, and a new record on Simple Bench!

Topics: Post-training Dominance · ARC-AGI 2 Caveat · Simple Bench Record · Hallucination Caveat · Model Card · Exponential Coming

The Two Best AI Models/Enemies Just Got Released Simultaneously

81K views · 9mo ago · 19:50

The two models that you will hear discussed for at least the next two months - Claude Opus 4.6 and GPT 5.3 Codex - just got released within 26 mins or each other. The full breakdown of around 250 pages of reports, with just the most interest moments, from the battle of which is best, Claude personhood, the surprising misbehaviour of Opus 4.6, and much more

Topics: Self-improvement? · Knowledge Work · Overly agentic behaviour · Who Shouldn’t Use Claude Opus · Step-change? · Claude’s ‘Personhood’

📄 Mentioned papers: Browse Comp · Terminal Bench 2

Claude AI Co-founder Publishes 4 Big Claims about Near Future: Breakdown

73K views · 9mo ago · 22:13

Anthropic's CEO, who has consistently predicted transformative AI will arrive before 2030, recently published a nearly 20,000-word essay outlining his vision of where AI is heading. The video gives you the highlights. The essay argues that scaling and recursion will advance AI from coding automation to full engineering automation, while warning of economic displacement within 1-2 years and China's trajectory toward AI-enabled totalitarianism.

Topics: Scaling to software engineers · Permanent Underclass · Totalitarian Nightmares · Collection of Personas

▶️ Mentioned videos: Video on youtube.com · Demis Hassabis Counter

📄 Mentioned papers: Societies of Thought, Google DeepMind Paper

📰 Mentioned news: Physics Prediction · Axios · Anthropic LessWrong

Anthropic: Our AI just created a tool that can ‘automate all white collar work’, Me:

106K views · 10mo ago · 19:03

A new tool, with code written only by AI, has gone omega-viral: Claude Cowork. But is the hype justified? What do the stats say on productivity? Where is the truth in a sea of noise? What is truth? Can we handle the truth? Where's Nemo?

Topics: Claude Cowork · Productivity Speed-up + jobs · Comparing Models · Brittle AI Paper

▶️ Mentioned videos: Douglas Interview

📄 Mentioned papers: Illusion of Insight · Entropy Exploration · ProRL

📰 Mentioned news: Amodei Prediction

What the Freakiness of 2025 in AI Tells Us About 2026

126K views · 10mo ago · 33:27

It’s probably not possible to satisfactorily condense a 12 month’s worth of weird progress in AI, as well as predictions for the year to come, into one video. But I’m gonna try anyway because it has been a very strange time.

Topics: Reasoning Models ... and limits · A playable world · Realism · AI Slop gone mainstream · DolphinGemma · Public Mood

▶️ Mentioned videos: Video on youtube.com · AI Slop · Altman Interview

📄 Mentioned papers: Reasoning Trade-offs · METR Time Horizon · LLM Brainrot · Emotional Quotient

📰 Mentioned news: Article on shash42.substack.com · GPT-5 - Altman phd in everything · Article on theguardian.com · Benchmark Gaming

Gemini Exponential, Demis Hassabis' ‘Proto-AGI’ coming, but …

90K views · 10mo ago · 20:00

The condensed highlights of hours of AI lab leader interviews, last-48-hour model releases, Gemini 3 Flash insights (plus it’s hidden flaw), Hassabis’ ‘proto-AGI’ and much more... Chapters Gemini 3 Flash Hassabis Interview Legg Interview Pre-training Lead Interview Altman Interview Brockman Video Post-Training Reveal

Topics: Results · But... the Flaw · So Benchmarks are fake? No · Spatial Reasoning + Hassabis · Proto-AGI · Minimal AGI

▶️ Mentioned videos: Hassabis Interview · Legg Interview · Pre-training Lead Interview · Altman Interview

📄 Mentioned papers: arXiv 2511.13029

📰 Mentioned news: OpenAI Compute Trend · OpenAI Valuation · Indian Data

GPT 5.2: OpenAI Strikes Back

89K views · 11mo ago · 17:42

Full GPT-5.2 breakdown - did OpenAI reclaim the crown? A story of tokens, time and cost, plus 9 details you wouldn’t get just from reading the headlines. Release Page GPT 5.2 Benchmark Comparison Charxiv GDPval My vid Kilpatrick

Topics: Better than Human @ Professional Tasks? · Test time Compute · Benchmark Selection · Simple Results + council comparison · Long Context · Self-Improvement

▶️ Mentioned videos: vid

📄 Mentioned papers: GDPval

📰 Mentioned news: Model in Year

You Are Being Told Contradictory Things About AI

76K views · 11mo ago · 20:16

With headlines of an imminent job apocalypse, code red for ChatGPT and recursive self-improvement, at the same time as Anthropic's CEO yesterday saying we know how to scale to AGI, and Gemini 3 DeepThink out today, it is easy to get lost among the narratives and counter-narratives. So here are both, plus the facts behind them, for you to decide.

Topics: Job Apocalypse? · Scaling to AGI · Recursive Self-Improvement Needed, or Not · OpenAI Code Red vs Gemini 3 DeepThink vs Claude Opus 4.5 · DeepSeek Speciale vs Mistral Large v3 · Claude Soul Document

▶️ Mentioned videos: Amodei on Scaling

📄 Mentioned papers: DeepSeek Paper

📰 Mentioned news: Guardian Interview · Article on cnbc.com · Claude Soul Document · Economist Worker Usage of GenAI

Nano Banana Pro: But Did You Catch These 10 Details?

61K views · 11mo ago · 14:56

Guest: Nano Banana Pro

Nano Banana Pro is no incremental upgrade. From double exposures to almost (!) perfect infographics, comic strips, to pricing, benchmarks and more, I go through 10 facets of the new, model you won’t get just from reading the headlines. Definitely worth watching before you use it for work...

Gemini 3 Pro: Breakdown

119K views · 11mo ago · 21:43

Gemini 3 Pro is out, and records fell like snowflakes in Svalbard. No long description, chapters or links today, huge technical difficulties, including with audio, so just want to publish asap. Podcast

Is GPT-5.1 Really an Upgrade? But Models Can Auto-Hack Govts, so … there’s that

62K views · 12mo ago · 18:27

A lot just got released in the last 36 hours, and it will all affect hundreds of millions of people. 10 details you would miss if you just read the headlines, from GPT 5.1 regressions, to how Claude hacked Govt Agencies, to SIMA 2, and Musical Turing Tests.

Topics: GPT 5.1 Smarter? · Some Regressions · Sycophancy? · Claude Auto-Hacking · Jailbreaking through Granularity · This Will be Re-used

📰 Mentioned news: Reuters Music

Bubble or No Bubble, AI Keeps Progressing (ft. Relentless Learning + Introspection)

61K views · 1y ago · 12:54

Guest: Relentless Learning

Don’t let headlines about bubbles distract you from the real avenues of progress being explored in AI every week, including what had been thought to be a long-term blocker - continual learning (learning on the fly). This, plus models introspecting (hesitate before you berate), Nano Banana 2 possibly spotted, Chinese imagen and more.

Topics: Continual Learning (Nested Learning / HOPE) · Introspection · Image-Gen Progress

📄 Mentioned papers: Original Titans Paper

📰 Mentioned news: Siri News

Did you miss these 2 AI stories? A *Real* LLM-crafted Breakthrough + Continual Learning Blocked?

58K views · 1y ago · 14:14

While compute-spend focuses on cash over cleverness, it can seem like the AI IQ plateau will last forever. But here’s a breakthrough for science, led by an LLM. And not even the latest or greatest one! I’ll go over CS2-Scale, then the Definition of AGI paper, a key continual learning quote, and Sora 2 answering Math questions...

Topics: C2S · OpenAI not too far behind (Simple, Codex) · A Definition of AGI? · OpenAI Researcher on Continual Learning Problems · Sora 2 can answer math qs

▶️ Mentioned videos: Jerry Tworek Interview

📄 Mentioned papers: Paper on biorxiv.org

Sora 2 - It will only get more realistic from here

59K views · 1y ago · 15:44

Sora 2 - the start of the infinite slop-feed or a key step to a generalist agent? Better than VEO 3 or over-hyped? I bring out 6 details you may have missed, contrast the announcement to Periodic Labs and even squeeze in some Claude Sonnet 4.5 analysis. Maybe I should make my videos longer...

Topics: Two models? · Rollout Details · Versus Sora 1 / Veo 3 · Sora App / Social Media · Masterplan · Generalist Agent? Periodic Labs

📰 Mentioned news: Altman on Regulations · OpenAI Profit · Article on nytimes.com

OpenAI Tests if GPT-5 Can Automate Your Job - 4 Unexpected Findings

67K views · 1y ago · 14:07

An OpenAI report released in the last 24 hours is the best look we have as to whether 2025 AI can automate your job. I’ll go through 4 unexpected findings, from which model is best at what, to practical tips and massive caveats. Plus UFC robots, radiologist essay, don’t trust videos and the blockers to the singularity.

Topics: OpenAI Report Summary · Tipping Point Speed-up · Better than Industry Experts? · Big Caveat · Karpathy and the Radiologist Analogy

ChatGPT Can Now Call the Cops, but 'Wait till 2100 for Full Job Impact' - Altman

48K views · 1y ago · 11:32

Sam Altman, CEO of OpenAI, announced a set of new ‘protections’ and ‘privileges’ for ChatGPT users, requiring a significant amount of trust from users. From predicting your age based on your chat to calling law enforcement if you are at risk of harm, to allowing non-minors to flirt. But amidst all of these announcements, there are interview snippets you may have missed, as Altman dramatically revises his predictions of AI impact on jobs. Plus a Hassbis backtrack to boot.

Topics: Calling the Cops? Age Predictions · Flirting and Privacy + Policymakers · Altman 70% of Jobs Gone, or 75 Years? · Hallucinations Paper · Demis Hassabis About Turn

▶️ Mentioned videos: Carlsen Interview · vs Senate Testimony (70 Jobs) · Hassbis Quote 1 · vs Quote 2

An ‘AI Bubble’? What Altman Actually said, the Facts and Nano Banana

58K views · 1y ago · 18:55

Wait, why did Sam Altman say AI was in a bubble? Or did he? Is it? 8 points for you to consider, before we all get distracted by Nano Banana. Bubble about to burst Nano Banana McKinsey Report Revenue MIT Report Safe Superintelligence

Topics: Sam Altman Clarification · Media Calls a Bubble (for the tenth time) · MIT and McKinsey Analysed · Incremental Progress Deceptive · Reasoning Breakthroughs · CEOs might not know their products

📄 Mentioned papers: Dawn of Reasoning? · arXiv 2403.04121 · Turing Test

📰 Mentioned news: Bubble about to burst · Revenue · Safe Superintelligence · Thinking Machines Lab

GPT-5 has Arrived

163K views · 1y ago · 15:02

GPT-5 will change how hundreds of millions of people use AI. Yes, you might have to forgive the chart crimes, the underwhelming livestream and Altman hype... But it’s a good model. I have read the 50 page system card in full, have the benchmark scores, coding tests, and things you might have missed.

▶️ Mentioned videos: Livestream

Genie 3: The World Becomes Playable (DeepMind)

201K views · 1y ago · 11:55

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.

Topics: Background and Access · Caveats · Demo

▶️ Mentioned videos: Genie 2 Coverage · TED Talk Roblox

How Not to Read a Headline on AI (ft. new Olympiad Gold, GPT-5 …)

84K views · 1y ago · 17:20

GPT-5 did what? OpenAI ahead of Google? There are 9 ways to misread the headlines of the last 48 hours, so this video is here to tell you what happened, sans sizzle. It’s been a fairly momentous last few days, so let’s dive in to the International Math Olympiad Gold, GPT-5 alpha release, whether mathematicians are out of jobs, and the white collar impact by year’s end.

Topics: AI Beat Mathematicians? · OPENAI vs GOOGLE · Irrelevant to Jobs or ... · White-collar jobs gone? · AI is Plateauing? · We Don’t Know the Details...

▶️ Mentioned videos: Livestream

📰 Mentioned news: 2k per month

Grok 4 - 10 New Things to Know

180K views · 1y ago · 11:44

Grok 4 is here, but did you know these 10 things about the new model? From benchmark caveats to soloing science, $300 a month secrets to Grok 5 promises, here's 10 new things to know in just under 12 minutes. Livestream Gemini DeepThink

Topics: Benchmark Results · Benchmark Caveats · ARC-AGI 2 · SimpleBench · ‘Humanity’s Last Exam’ · SuperGrok Heavy Price

▶️ Mentioned videos: Livestream · SmartGPT · Video on youtube.com · Video on youtube.com

📰 Mentioned news: Pollution · Burn Rate

When Will AI Models Blackmail You, and Why?

111K views · 1y ago · 26:20

In the last few days Anthropic have released an impressive honest account of how all models blackmail, no matter what goal they have, and despite prompt warnings, and other preventions. But do these models want this? Report 30 Page Appendices Announcement OpenAI Files Grok 4 News Claude 4 Report Card New Apollo Research Interesting Reflections

Topics: What prompts blackmail? · Blackmail walkthrough · ‘American interests’ · Inherent desire? · Switching Goals · Murder

Apple’s ‘AI Can’t Reason’ Claim Seen By 13M+, What You Need to Know

102K views · 1y ago · 14:01

What to make of those headlines that AI can’t reason, seen by tens of millions? I cover the Apple paper in layman’s terms, what it means and doesn’t mean, and what’s next. Plus o3-pro and whether it is my current most-recommended model.

Topics: Viral Post + Headlines · Apple Paper Analysis · But they do Hallucinate · Not Supercomputers · o3 Pro and Recommendations

▶️ Mentioned videos: Alpha Evolve Video

📰 Mentioned news: Guardian Article · Marcus · AI Job Headlines · Article on axios.com

AI Accelerates: New Gemini Model + AI Unemployment Stories Analysed

96K views · 1y ago · 16:50

Guest: AI Accelerates

There’s a new best language model, so let’s go through the up and downs of Gemini 2.5 Pro 06-05. Record-breaking common-sense, but dumb mistakes remain. And it’s not even their best model, which remains behind the scenes - Gemini 2.5 Ultra. Plus Sundar Pichai’s AGI date and an analysis of whether the current AI unemployment headlines are justified, and Elevenlabs v3.

Topics: Gemini 2.5 Ultra · Benchmarks · AGI Date and Meaning Pichai · Jobs and AI Unemployment Fears · Elevenlabs v3

▶️ Mentioned videos: Sundar Pichai Fridman · Elevenlabs v3

📰 Mentioned news: White Collar Bloodbath · Article on fortune.com · Article on nytimes.com · Article on nytimes.com

Claude 4: Full 120 Page Breakdown … Is it the Best New Model?

99K views · 1y ago · 19:05

Cursor Says State of the Art for Coding Benchmarks Podcast

Topics: 3 Quick Controversies · Benchmark Results · 120 page Card 20 Highlights · Coding Test · Model Welfare and Spiritual Bliss · ASL-3

Google Takes No Prisoners Amid Torrent of AI Announcements

99K views · 1y ago · 17:08

Google just announced at least 12 things that are each worthy of a video, but here are the top I/O highlights. From Veo 3 to Deep Research now being useable, Deep Think breaking records to Gemini Diffusion, Gemini 2.5 Flash changing how AI is priced and GemmaVerse, SynthID Detector and Imagen 4. And even this intro is missing other announcements covered in the vid! And yes, they’ll be plenty of Veo 3 clips to enjoy...

Topics: Veo 3 · Gemini 2.5 Flash · Universal Assistant · Usage Skyrockets + OpenAI dig · Gemini Pro Deep Think · Overviews and AI Mode

▶️ Mentioned videos: Event · SignGemma (See end of this vid)