Gregory Gromov
@ntvll
Followers
751
Following
10K
Media
12K
Statuses
42K
MS Avionics, Autoformalization, Autonomous Vehicles, Road Safety Metrics
USA
Joined December 2013
The idea of the intellectual immortality of generations is knocking at the door of the next stage of the AI revolution. To give a first impression of what we are talking about, let me recall a counter-intuitive statistic on road safety by age groups: the safest drivers, by
2
0
14
Such a combination of creative autonomy and real-time translation of theoretical ideas into large-scale operational systems … as experienced by Ilya Sutskever in collaboration with Sam Altman and Greg Brockman … is uncommon and difficult to replicate in other institutional
0
0
0
Silicon Valley’s impact can’t be reduced to the 2-3 decades short list of inventions What it repeatedly produced were platform revolutions: • microprocessors revolution • personal & mobile computing • the Internet & cloud ecosystem • and, most recently, the AI revolution
0
0
0
Recent shifts in the Apple–OpenAI–Google relationship have noticeably changed the balance of expectations between AI-based search and legacy search ecosystems, while the rumors about OpenAI search agent have intensified. Below are the arguments often cited in that context.
0
0
0
“In 15 days, 3 Erdős problems were solved (GPT-5.2 Pro).” Meanwhile, Gemini is eating GPT’s breakfast, lunch, and dinner, while Apple pushed GPT out of the core of its ecosystem … all for the same reason: Sam Altman, Ilya Sutskever, and Greg Brockman led OpenAI as
We're only 15 days into the new year. - In 15 days, 3 Erdos problems were solved (GPT-5.2 Pro), - Grok 4.20 (pre-access) found a new Bellman function for one of the problems Prof. Vanisvili was working on, - The CEO of Cursor used coordinated hundreds of GPT-5.2 agents to
0
0
0
Any software engineer knows that the hardest part of the job is understanding what a client actually needs versus what they say they want. Everything else is easy by comparison; even Al models can handle the coding part now. Ultimately, software engineers have little reason to
A wife tells her programmer husband: “Go to the store and buy a gallon of milk. If they have eggs, get six.” He comes back with six gallons of milk. When she asks why, he replies: “They had eggs".
0
0
1
LLM era prehistory: 1997 (Deep Blue): The super-fast computer defeated a human — chess champion — but failed to become a daily companion. OpenAl became a pioneer of the LLM stage of the Al revolution: 2023 (ChatGPT): The 'co-thinking' mode opened the Al era for a billion OpenAl
0
0
0
Apple prefers not to "put all its eggs in one basket." Following the integration of OpenAI’s ChatGPT … a new report from CNN confirms that “Apple teams up with Google Gemini for AI-powered Siri.” Apple becomes the AI race ultimate kingmaker. By splitting its business
ChatGPT was going to be ultra-integrated into Windows (and then that kind of just fizzled out) Then ChatGPT was going to be ultra-integrated into Apple (that kind of just fizzled out) What is the common factor here?
0
0
0
The next most impressive AI achievement after IBM computer won chess championship against Gary Kasparov :
AI just achieved a perfect score on the hardest math competition in the world. The Putnam has 12 problems x 10 points. The highest score last year was 90. The median? 0. Axiom's AI prover in Lean scored 120/120 and just shared all the solutions. Huge milestone for AI in Math.
0
0
0
‘If you are going through hell, keep going." - Churchill What happened in 2023-2024: ChatGPT triggered the LLM stage of the Al revolution, while Tesla's annual sales growth stalled to zero. Musk reasonably concluded that the era of the EV frontier was over — centuries long
Tesla's 2025 new car deliveries have reduced its compound annual growth rate over the past 12 years to *an eleven year-low* of ONLY... 43%. 😂 Still pretty good! $TSLA
0
0
0
Claude is qradually materializing the AUTOFORMALIZATION-1986 basic concept … by building user’s personal knowledge bases over time of maintaining project state, preserving effectively unlimited context, while openAI keeps bragging about model 5.2 next success in math
@gdb This is exactly why GPT degrades so quickly. Instead of refining the 'co-thinking' that made early OpenAl models successful, the focus has shifted to rigid, formalized tasks. While Gemini and Grok are expanding user memory, GPT 5.2 is cannibalizing it. Long, fruitful threads
0
0
0
BYD dethroned Tesla as world EV leader:
BYD outsold Tesla by over 600,000 electric vehicles in 2025. Final tally: - BYD: 2.25 million EVs - Tesla: 1.64 million EVs Tesla's sales fell 9% year-over-year. BYD's grew 28%. First time in the modern EV era that Tesla isn't the global leader.
0
0
0
Meanwhile, the 5.2 —next model —almost completely https://t.co/AkP28rV3we destroyed all of the above expectations
@gdb This is exactly why GPT degrades so quickly. Instead of refining the 'co-thinking' that made early OpenAl models successful, the focus has shifted to rigid, formalized tasks. While Gemini and Grok are expanding user memory, GPT 5.2 is cannibalizing it. Long, fruitful threads
0
0
0
Google Search provides a perfect example of 'memory of the future' — most popular results on the first page are almost completely occupied by PPC ads and SEO products.
Over time, all LLMs will pretty much collapse into simply recycling each other's outputs. The entire internet will be a bunch of AI-generated content. All new AI models will train on this content and regurgitate it back to us in an endless cycle. The end result - A giant
0
0
0
Generally, there are only three companies with a decade or more of sustained experience applying AI to the physical world. Tesla—despite its well-known road-safety issues—is one of them. As Yann LeCun emphasized in his recent Paris remarks, most of the apparent progress in AI
@Teslaconomics You’re right. The actual time from when FSD sort of works to where it is much safer than a human is several years. The legacy car companies won’t design the cameras and AI computers into their cars at scale until several years after that. So this is maybe a competitive
0
0
0
The most significant result of the first two years of the LLM stage of the Al revolution … was the increased productivity of software engineers working at companies that produce LLMs … ”but still fails at basic human tasks”
First 2 years of the race to AGI — the top achievement: ”Claude can code as well as some of the anthropic best engineers, but still fails at basic human tasks” Assembler, Fortran, … Claude — done.
0
0
0