Person
Steve Newman
Steve Newman co-created Writely, the web word processor that became Google Docs, and now writes Second Thoughts, a newsletter that brings an engineer's patience to the biggest questions about AI — what the technology can actually do, how fast it is improving, and what would count as evidence either way. He also co-founded the Golden Gate Institute for AI.
Commentary by Steve Newman
From the commentary rail — every link leaves the site for the original piece.
- 1 September 2026 · Second Thoughts14 Reasons Robotics is HardAnd why you should ignore demo videos
- 21 August 2026 · Second ThoughtsCybersecurity: Let’s Play to WinStop expecting perfection and start making it unnecessary
- 20 July 2026 · Second ThoughtsAnecdotes Everywhere, Evidence Almost NowhereThe state of AI in mid 2026, part 3
- 30 June 2026 · Second ThoughtsNever Trust A NumberThey're shortcuts to understanding, and there are no shortcuts
- 19 May 2026 · Second ThoughtsFrom Compute Overhang to Compute CrunchThe state of AI in Q2 2026, part 2
- 7 May 2026 · Second ThoughtsIs AI 2027 Coming True?The state of AI in Q2 2026, part 1
- 17 April 2026 · Second ThoughtsThe Centaur EraThe question isn't what AI can do, it's what you can do with AI
- 2 March 2026 · Second Thoughts45 Thoughts About AgentsThe layer of the AI stack that evolves fastest – and may have the most impact
- 10 February 2026 · Second ThoughtsWe Just Got a Peek at How Crazy a World With AI Agents May BeClawdbot and Moltbook: Cosplaying a Future of Independent AIs
- 14 January 2026 · Second ThoughtsDiscarding the Shaft-and-Belt Model of Software DevelopmentAI enables – and benefits from – a move from mega-projects to artisanal solutions
- 6 January 2026 · Second ThoughtsSoftware Too Cheap to MeterPersonalized solutions may replace one-size-fits-all applications
- 19 December 2025 · Second ThoughtsPresenting the Case That the Future Will Be UnrecognizableWe should prepare for the possibility of profound change
- 20 November 2025 · Second ThoughtsReflections on The CurveA compilation
- 19 November 2025 · Second ThoughtsHyperproductivity: The Next Stage of AI?A glimpse at an astonishing, exhilarating, exhausting new style of work
- 4 November 2025 · Second ThoughtsA Project Is Not a Bundle of TasksCurrent AIs struggle to create a whole that exceeds the sum of its parts
- 29 October 2025 · Second ThoughtsWhat I Saw Around The CurveNotes from the near-future of AI
- 24 September 2025 · Second ThoughtsHuman Drivers Will Kill 11 People While You Read ThisLet’s not obstruct the technology that could have saved them
- 10 September 2025 · Second ThoughtsWe’re In the Windows 95 Era of AI Agent SecurityHeaded for widespread usage, and insecure by design
- 4 September 2025 · Second ThoughtsGPT-5: The Case of the Missing AgentProgress Everywhere Except In The Real World
- 14 August 2025 · Second Thoughts35 Thoughts About AGI and 1 About GPT-5Such As: How do Teenagers Learn to Drive 10,000x Faster Than Waymo?
- 10 July 2025 · Second ThoughtsNot So Fast: AI Coding Tools Can Actually Reduce ProductivityStudy Shows That Even Experienced Developers Dramatically Overestimate Gains
- 30 May 2025 · Second ThoughtsLab vs. Life: Dissecting “AI as Normal Technology”What if AGI Can't Be Developed in an Ivory Tower?
- 21 May 2025 · Second ThoughtsWhen Decades Become Days: Dissecting AI 2027The Four Requirements for a Short AI Timeline
- 24 April 2025 · Second ThoughtsFinally, A Way to Measure AI Progress; Everyone's Misreading ItThat METR Study Doesn’t Say "AGI in 5 Years"
- 27 February 2025 · Second ThoughtsIf AGI Means Everything People Do... What is it That People Do?And Why Are Today’s "PhD" AIs So Hard To Apply To Everyday Tasks?
- 18 February 2025 · Second ThoughtsWe're Finding Out What Humans are Bad AtAI Advances Fastest When We Find Unnatural Ways of Doing Things
- 30 January 2025 · Second ThoughtsChina’s DeepSeek Adds a Weird New Data Point to The AI RaceV3 and R1 are Impressive Work, With Many Implications – but not "China Has Caught Up"
- 6 January 2025 · Second ThoughtsAre We on the Brink of AGI?A Tale of Two Timelines
- 3 January 2025 · Second ThoughtsThe Important Thing About AGI is the Impact, Not the NameReality Doesn't Care How We Interpret the Words "General Intelligence"
- 27 December 2024 · Second ThoughtsLLMs Fight With Both Hands Tied Behind Their BackWe haven't yet given them access to knowledge-in-the-world
- 24 December 2024 · Second ThoughtsThe Black Spatula Project: Day FiveOff to a Roaring Start
- 18 December 2024 · Second ThoughtsThe Black Spatula ProjectFixing flawed scientific papers, 5000 tokens at a time
- 16 December 2024 · Second ThoughtsThe Future Is Already Here, It’s Just Not Evenly DistributedYou Can Awe Some of the People Some of the Time
- 20 November 2024 · Second ThoughtsWhat Are the Real Questions in AI?It's hard to have a constructive discussion if we're not talking about the same thing
- 12 November 2024 · Second ThoughtsAI is Racing Forward – on a Very Long RoadProgress Hasn’t Stalled; But AGI Is Not Yet Near
- 26 October 2024 · Second ThoughtsTo Change the World, Set a Bold Target: Moore's Law as Self-Fulfilling ProphecyAll progress depends on the unreasonable forecast
- 16 October 2024 · Second ThoughtsNo, A Bot Didn't Just Make $150M in CryptoBeware the Startling Anecdote
- 13 October 2024 · Second ThoughtsLearning to Explore: AlphaProof and o1 Show The Path to AI CreativityAn Impressive Start, But Many Hurdles Lie Ahead
- 2 October 2024 · Second ThoughtsAI Creativity Is A Question of Quality, Not NoveltyIt Actually Matters How Well The Bear Dances
- 17 September 2024 · Second ThoughtsWhat It’s Like To Solve a Math Olympiad ProblemAnd What This Tells Us About Creativity, Reasoning, And AI
- 13 September 2024 · Second ThoughtsAI Could Break Things; Let's Use It As a Wakeup Call To Make Them StrongerStop Treating AI Policy Tradeoffs As Zero-Sum
- 30 August 2024 · Second ThoughtsRather Than Arguing About What We Don't Know, Let's Work Together To Find OutA Lesson From The SB 1047 Debate
- 6 June 2024 · Second ThoughtsA Response to "Situational Awareness"I think the scenario is wrong; let's come up with a plan that works either way
- 31 May 2024 · Second ThoughtsGrounding the Conversation About AIOrganizing The (AI) World’s Disagreements And Making Them Universally Accessible and Constructive
- 2 May 2024 · Second ThoughtsOne Conversation is Worth a Thousand Angry TakesMake The Internet Better Using This One Weird Trick
- 25 April 2024 · Second ThoughtsWhat To Expect When You’re Expecting GPT-5Where Will AI Go From Here?
- 17 March 2024 · Second ThoughtsI Don't See How Comparative Advantage Applies In a World of Strong AIEconomic models are based on simplifying assumptions that may not hold in an AI-saturated future
- 19 February 2024 · Second ThoughtsWhy Are LLMs So Gullible?It’s Because They’re Naive and Constantly Confused
- 25 January 2024 · Second ThoughtsAnatomy of a HackPractical hacks are complicated. That means complicated systems are unsafe.
- 17 January 2024 · Second ThoughtsWhat is “Prompt Injection”, And Why Does It Fool Chatbots?LLMs See Words, But Not Context
- 15 December 2023 · Second ThoughtsWould We Really Shut Down A Misbehaving AI?Ask The Old OpenAI Board How Things Went When They Tried To Shut Down Sam Altman
- 20 November 2023 · Second ThoughtsToward Better AI MilestonesHow to get off the treadmill of constantly shifting goalposts, and define tests that will tell us when AI capabilities get serious
- 27 September 2023 · Second ThoughtsWhen Will AIs Acquire Insight?How Could We Even Train Them For That?
- 19 September 2023 · Second ThoughtsThe AI Explosion Might Never HappenAs AI Learns To Improve Itself, Improvements Will Become Harder To Find
- 12 September 2023 · Second ThoughtsIntermediate SuperintelligenceGodlike AI, if it happens at all, won't happen overnight
- 1 September 2023 · Second ThoughtsWhen We Forecast AGI, Do We Mean In The Lab Or In The Field?The Onset of AGI Will Play Out Over Decades
- 18 August 2023 · Second ThoughtsThe AI Progress ParadoxReceding Horizons and Misleading Milestones
- 9 August 2023 · Second ThoughtsWhy We Won't Achieve AGI Until Memory Is A Core Architectural ComponentThe Hard Part of Thinking Is Knowing What To Think About
- 18 July 2023 · Second ThoughtsProcess vs. Product: Why We Are Not Yet On The Cusp Of AGIYou Can't Learn The Journey By Observing The Destination
- 11 July 2023 · Second ThoughtsWe Need to Recognize How Profoundly Different The AGI Future Will BeWe are shaping AI; soon it will also be shaping us
- 17 June 2023 · Second ThoughtsWhat Will AI Do For Us In The Near Term?The more time I spent writing this post, the more optimistic I became
- 29 May 2023 · Second ThoughtsTo Address AI Risks, Draw Lessons From Climate ChangeWe can't solve the problem today; we *can* start establishing the conditions for it to be solved.
- 24 May 2023 · Second ThoughtsA Unified Theory of AI RiskNear-term risks aren't a distraction from extinction scenarios, they're the practice exam
- 4 May 2023 · Second ThoughtsGet Ready For AI To Outdo Us At EverythingWe have time to prepare; step one is acknowledging what we're preparing for
- 30 April 2023 · Second ThoughtsBeyond the Turing TestThe key question about AI is no longer "can it think", but rather "can it hold down a job"?
- 25 April 2023 · Second ThoughtsAn Intuitive Explanation of Large Language ModelsInstead of diving into the math, I explain *why* they're built as "predict the next word" engines, and present a theory for why they make conceptual errors.
- 20 April 2023 · Second ThoughtsI’m a Senior Software Engineer. What Will It Take For An AI To Do My Job?Exploring the gaps between current LLMs and true general intelligence
- 13 April 2023 · Second ThoughtsGPT-4 Doesn't Figure Things Out, It Already Knows ThemYes, It's a Stochastic Parrot, But Most Of The Time You Are Too, And It's Memorized a Lot More More Than You Have