Technologist & AI Strategist | 25 Yrs in Tech | NYU Lecturer & AI Mentor | Writer at The AI Spectator | Author of The Tony Hawk Paradox
New York, NY
Joined March 2009
- Tweets9.1K
- Following875
- Followers2.2K
- Likes4.4K
Pinned Tweet
In 1999, the original @tonyhawk Pro Skater game included tricks the designers built purely as fantasy. As Tony Hawk himself put it in April 2025: "We put that in there because it was fantastical. It was like, 'this would be amazing, but you can't do it in real life.' Now they do that."
The 1080-degree spin, coded into the game as pure imagination, was first landed in 2012 by a twelve-year-old who grew up playing it. The game came first. Reality came second.
My new book argues this isn't a fluke. It's a pattern that shows up everywhere once you look.
The world's leading AI researcher, Nobel laureate @demishassabis of @GoogleDeepMind , says it plainly: "Games are always just the proving ground for our algorithms." DeepMind masters games first, then applies those breakthroughs to protein folding and beyond. @BostonDynamics trains robots across thousands of digital simulations before a single physical attempt. Athletic records that stood as absolute limits fell and became routine. IQ scores climbed for a century.
Simulation first. Reality second. Again and again. Sign up below to be notified of the release date and for a chance to receive a signed copy.
tonyhawkparadox.com/
Elon Musk wrote on October 29th that Neuralink's Blindsight vision implant is ready from an engineering standpoint and waits on regulatory approval, the position he stated in January.
No human Blindsight implant has been reported, and Neuralink's trial page shows no active Blindsight study. Cortical vision implants from other groups have produced spots of light that helped users find objects and identify some letters, which sets the baseline Blindsight will be measured against.
The same update covered speech restoration results from an ALS participant and a 72-hour engineering sprint. I discuss this and more in today's article.
Read the full story: davidborish.com/post/neurali…
Elon Musk wrote on October 29th that Neuralink's Blindsight vision implant is ready from an engineering standpoint and waits on regulatory approval, the position he stated in January.
A control policy that failed all 256 test episodes succeeded in all 1,024 after GPT-6 Astra edited 3.5% of its weights in one reply, with no tools.
MIT CSAIL researchers report that the same model identified what small networks were trained on from raw weights, wrote game-playing networks starting from zeros, and raised Qwen3.5-0.8B from 7.9% to 48.1% on MATH-500 by changing 0.016% of its weights. I discuss this and more in today's article.
Read the full article: davidborish.com/post/reading…
A control policy that failed all 256 test episodes succeeded in all 1,024 after GPT-6 Astra edited 3.5% of its weights in one reply, with no tools
The Biological Computing Co. released Mischief 1 on October 30, an image-to-video model built on Alibaba's Wan 2.2 that the company says runs five times faster at 80 percent lower cost, from $0.005 per second at 480p.
The Biological Computing Co. released Mischief 1 on October 30, an image-to-video model built on Alibaba's Wan 2.2 that the company says runs five times faster at 80 percent lower cost, from $0.005 per second at 480p.
The speedup comes from a small software adapter designed with recordings of lab-grown rat and human stem cell derived neurons, and no cells are involved when a clip is generated.
The speed claims are the company's own, and the published neuron research used a different model. The ethics objections to human-derived neurons come from the wider biocomputing field, and the reporting reviewed here turned up no named critic of this product.
Read the full article: davidborish.com/post/the-bio…
Love this idea!
An influencer is incentivized by attention, so their content must be optimized to be consumed and forgotten (cheap dopamine).
A personal brand is incentivized by money, so their content must be optimized to move you down their sales funnel.
An artisan is incentivized by the satisfaction that comes from the quality of their work. They need attention and money, but for the sole purpose of spending more time on work they enjoy and the impact it has.
Become an artisan.
A DeepMind essay published today models AI's economic effects around how well agents work together.
Alex Imas, Andrew Koh and Shane Legg describe four capability regimes and say wage and job-role changes will show up well before GDP or industry concentration moves.
In their framework, a firm whose members each add to its output as it grows would break the usual limit on company size. They also say today's agents coordinate poorly.
Read the full article: davidborish.com/post/the-eco…
Anthropic has published results from drug development partnerships with six organizations, including a CAR T-cell component 70% shorter than the standard version that produced roughly 10 times as many receptors in lab tests.
Anthropic has published results from drug development partnerships with six organizations, including a CAR T-cell component 70% shorter than the standard version that produced roughly 10 times as many receptors in lab tests.
A second partner saw Claude generate about 35,000 antibody designs in a month. A third cut manufacturing document review from over 25 hours to under three. The work is mostly preclinical, and every figure comes from Anthropic’s own account.
Read the full article: davidborish.com/post/anthrop…
Google has released a research preview of Genie 4, a world model that generates video, ambient sound and spoken dialogue together while users steer the scene with text prompts.
Google has released a research preview of Genie 4, a world model that generates video, ambient sound and spoken dialogue together while users steer the scene with text prompts.
A second Gemini model works inside the loop, checking each frame against a step-by-step plan. Google’s announcement includes demonstration prompts, tester feedback and a candid list of limitations, but no benchmarks, latency figures or session lengths. Access is limited to trusted testers for now.
Read the full article: davidborish.com/post/genie-4…
OpenAI reported three more findings in Math, Computer Science, and Physics on October 26.
They claimed that its models produced a disproof of the Exponential Time Hypothesis, a proof of the local smoothing conjecture in every dimension, and existence proofs for Yang-Mills theory and quantum chromodynamics.
OpenAI reports Lean formalizations for the first two and for the Yang-Mills proof. Outside groups led by Russell Impagliazzo, Ramamohan Paturi, Ryan Williams and David Gross will review the work, and none of it has been peer reviewed. The groups chose different release terms, and the physics group has withheld the AI-produced materials for now. I discuss this and more in today's article.
Read the full article: davidborish.com/post/openai-…
A developer used Claude Opus 5.5 to build open-source re-implementations of Photoshop, Word, and Excel in a few weeks, and McKinsey found that 32% of organizations have already declined a software purchases.
A developer used Claude Opus 5.5 to build open-source reimplementations of Photoshop, Word, and Excel in a few weeks, and McKinsey found that 32% of organizations have already declined a software purchase because they could build the tool themselves with coding agents.
The new edition of The Open-Prem Inflection Point V4, out today extends the framework from AI models to application software.
It covers which software companies should build, adopt, or keep buying, why the license of the starting code decides what they can keep private, and where maintenance and governance slow these projects down.
I would love to hear thoughts from my colleagues who have been tracking #OpenPrem from V1 Matt Konwiser Sudeep Kesh Akanimo Udo Steve Wylie Asher Lohman Ash Dhupar
Read the full article: davidborish.com/post/the-ope…
David Borish retweeted
The website for my book redone with Opus 5.5. The initial version (made with Opus 4.8) required a lot of work to make a distinctive-looking static site. Now Opus mostly did this alone. We have come a long way from generic ClaudeDesign in a very short time. co-existence.ai/
On October 21, Marathon Fusion posted that it had turned mercury into stable gold using fusion neutrons at ten times CERN's 2025 rate.
The data it showed is a gamma-ray spectrum consistent with the reaction the company proposed in 2025.
This article sets out what the evidence supports, what remains unpublished, and why the company's plasma centrifuge work stands on firmer ground.
Read the full article: davidborish.com/post/alchemy…
Last month, we demonstrated high-rate alchemy: turning mercury into real, stable gold, beating CERN’s record by a factor of 10.
Last week, we enriched isotopes in the world’s most powerful plasma centrifuge, beating the Soviet record that stood for 40 years.
Now, we’re building a fusion reactor in San Francisco to achieve record power density.