Wiredframe Radar logo

Category intelligence

Social Media Briefing — February 14, 2026

492 current items analyzed and ranked.

Executive synthesis

Social Media Summary

The AI community was electrified by OpenAI's announcement that GPT-5.2 derived a novel result in theoretical physics — a gluon interaction finding published as a preprint with IAS, Harvard, Cambridge, and Vanderbilt. Physicist Andy Strominger called it potentially unsolvable by humans alone, marking a watershed moment for AI-assisted scientific discovery.

  • Boris Cherny revealed Spotify's deep adoption of Claude Code: top developers haven't written code since December and ship features from Slack, signaling a cultural shift in enterprise engineering
  • Shane Legg (DeepMind co-founder) updated his AGI estimates to 50% by 2028 and 70% by 2030, while teasing upcoming AGI tests from Google
  • François Chollet pushed back, arguing AGI won't trigger an intelligence explosion — scientific progress is fundamentally linear, not exponential
  • Gary Marcus presented data-driven evidence that LLM hallucinations remain unsolved across law, science, and medicine
  • Andrew Ng shared insights from Sundance on AI-Hollywood tensions around IP licensing, job displacement, and creative control
  • Gary Marcus flagged that xAI appears to have no safety team based on an org chart Elon Musk shared publicly

Key Themes

GPT-5.2 Physics Discovery · 3AI Scientific Discovery (GPT-5.2 Physics) · 6AGI Timelines & Progress Dynamics · 12Claude Code Enterprise Adoption · 3Agentic Coding & Open Source Tools · 8LLM Hallucinations Debate · 9AI & Labor/Creative Industries · 9OpenAI Governance & Mission Evolution · 1OpenAI Product Updates · 5AI Agent Limitations & Reality Check · 3

Primary evidence

Top Ranked Signals

Social Twitter Feb 13

GPT-5.2 derived a novel result in theoretical physics, showing that a type of particle interaction m...

By @gdb

95 score
AI Analysis

OpenAI president Greg Brockman announces GPT-5.2 derived a novel result in theoretical physics, showing a type of particle interaction many physicists expected wouldn't occur can actually arise under specific conditions. Frames this as AI accelerating science.

GPT-5.2 derived a novel result in theoretical physics, showing that a type of particle interaction many physicists expected would not occur can in fact arise under specific conditions. There is great promise in the potential of AI to benefit people by accelerating science. t.co/B1zpYbKfcZ
ai_scientific_discoverygpt52_physicsopenai
Social Twitter Feb 13

GPT-5.2 derived a new result in theoretical physics. We’re releasing the result in a preprint with ...

By @OpenAI

92 score
AI Analysis

OpenAI announces GPT-5.2 derived a new result in theoretical physics — a preprint with IAS, Vanderbilt, Cambridge, and Harvard showing a gluon interaction previously thought not to occur can arise under specific conditions.

GPT-5.2 derived a new result in theoretical physics. We’re releasing the result in a preprint with researchers from @the_IAS, @VanderbiltU, @Cambridge_Uni, and @Harvard. It shows that a gluon interaction many physicists expected would not occur can arise under specific conditions. t.co/EAZhKWacsG
ai-scientific-discoveryphysicsopenaifrontier-modelsai-capabilities
Social Twitter Feb 13

“It is the first time I’ve seen AI solve a problem in my kind of theoretical physics that might not ...

By @gdb

88 score
AI Analysis

Greg Brockman quotes physicist Andy Strominger saying AI solved a problem in theoretical physics that 'might not have been solvable by humans' — follow-up to GPT-5.2 physics discovery announcement.

“It is the first time I’ve seen AI solve a problem in my kind of theoretical physics that might not have been solvable by humans.” — Andy Strominger
ai_scientific_discoverygpt52_physics
Social Twitter Feb 13

Love seeing how Spotify is shipping with Claude Code. Their best developers haven't written a sing...

By @bcherny

88 score
AI Analysis

Continuing from yesterday's Social coverage of Claude Code momentum, Boris Cherny shares how Spotify is using Claude Code: best developers haven't written code since December, fix bugs from phones, shipped 50+ features from Slack during commutes

Love seeing how Spotify is shipping with Claude Code. Their best developers haven't written a single line of code since December, they fix bugs from their phones, and they shipped 50+ features from Slack during morning commutes t.co/rYTVJBHE0s
claude-codeenterprise-adoptionagentic-codingdeveloper-productivityanthropic
Social Twitter Feb 13

I agree that AI testing is best thought of as a process and that what humans can typically do is a n...

By @ShaneLegg

85 score
AI Analysis

Shane Legg (DeepMind co-founder) gives updated AGI probability estimates: 50% chance by 2028, 70% by 2030. Agrees AGI testing should be a process and human-level performance is a natural minimum bar.

I agree that AI testing is best thought of as a process and that what humans can typically do is a natural minimal bar for AGI. Perhaps I'm bit more optimistic about the speed of progress: 50% chance by 2028, 70% by 2030.
agi_timelinesagi_evaluationgoogle_ai
Social Twitter Feb 13

Are LLM hallucinations basically solved, as a former Senior Policy Advisor at the White House, @dean...

By @GaryMarcus

82 score
AI Analysis

Marcus presents a detailed, data-driven case that LLM hallucinations are NOT solved: law (fake cases up from 100 to 900+), science ('hallucitation' coined), pharma (26-69% hallucination rates), 15%+ across models, OpenAI's own white paper acknowledging the problem, plus indirect measures like low task completion rates and poor enterprise ROI.

Are LLM hallucinations basically solved, as a former Senior Policy Advisor at the White House, @deanwball, told me below, based pure on anecdotal experience and without data? No. Instead, his post is symptomatic of how subjective finger-in-the-wind evaluations of AI often get things wrong. Here are some actual data. • Law: Incidents in which lawyers have gotten busted for using fake cases are way up; @DamienCharlotin’s database listed around 100 cases less than a year ago, and now has over
hallucinationsai_limitationsai_reliabilityenterprise_ai
Social Twitter Feb 13

The authors of the preprint realized this a year ago and sought to find the correct formula for inte...

By @OpenAI

82 score
AI Analysis

OpenAI details how GPT-5.2 simplified complex gluon interaction expressions that physicists couldn't simplify by hand, conjectured a general formula, and a separate scaffolded model independently derived and formally proved the same formula in ~12 hours.

The authors of the preprint realized this a year ago and sought to find the correct formula for interactions involving any number n of gluons, going up to n=6 by hand but obtaining complicated expressions that they sought to simplify, without success. GPT-5.2 simplified these expressions and then conjectured a simple formula for the general case. Next, a separate scaffolded internal OpenAI model spent roughly 12 hours reasoning through the problem, independently deriving the same formula and
ai-scientific-discoveryphysicsopenaiai-capabilitiesfrontier-models
Social Twitter Feb 13

I recently spoke at the Sundance Film Festival on a panel about AI. Sundance is an annual gathering ...

By @AndrewYNg

78 score
AI Analysis

Andrew Ng gives a detailed account of speaking at Sundance Film Festival about AI and Hollywood's concerns: IP/licensing, job displacement via unions, and feeling technology is forced upon them. Notes common ground exists but the path is unclear.

I recently spoke at the Sundance Film Festival on a panel about AI. Sundance is an annual gathering of filmmakers and movie buffs that serves as the premier showcase for independent films in the United States. Knowing that many people in Hollywood are extremely uncomfortable about AI, I decided to immerse myself for a day in this community to learn about their anxieties and build bridges. I’m grateful to Daniel Dae Kim @danieldaekim, an actor/producer/director I’ve come to respect deeply for hi
ai_creative_industryai_labor_impactai_ipai_adoption
Social Twitter Feb 13

“‘Safety is a dead org at xAI,’ he said. Looking at the restructured org chart Elon Musk shared on X...

By @GaryMarcus

75 score
AI Analysis

Gary Marcus reports that xAI has no safety team according to an org chart Elon Musk shared, calling a safety-indifferent zillionaire racing for AI dominance dangerous.

“‘Safety is a dead org at xAI,’ he said. Looking at the restructured org chart Elon Musk shared on X, there’s no mention of a safety team.” Last July I wrote about how Elon Musk had personally elevated my p(doom), from very low to low. This latest development is a natural extension of that fear. A zillionaire hell bent on winning the AI race and no longer interested in AI safety is not what the world needs.
ai_safetyxaiai_governance
Social Twitter Feb 13

I don't think the rise of AGI will lead to a sudden exponential explosion in AI capabilities. There ...

By @fchollet

75 score
AI Analysis

Chollet argues AGI won't lead to a sudden exponential explosion in capabilities because bottlenecks on capability improvements can't be lifted by horizontally scaling intelligence in silicon.

I don't think the rise of AGI will lead to a sudden exponential explosion in AI capabilities. There are bottlenecks on the sources of new capability improvements, and horizontally scaling intelligence in silicon (even by a massive factor) doesn't lift those bottlenecks.
agiintelligence_explosionai_progress_dynamics
Social Twitter Feb 13

If you have been using AI for a couple years, you really notice how good GPT-5.2/Opus 4.6/Gemini 3 r...

By @emollick

72 score
AI Analysis

Emollick reflects that GPT-5.2, Opus 4.6, and Gemini 3 are impressively smart and perceptive if you've used AI for a couple years, while still having a jagged frontier of abilities. Criticizes the focus on harnesses/CLIs over the models themselves.

If you have been using AI for a couple years, you really notice how good GPT-5.2/Opus 4.6/Gemini 3 really are. Forget the harnesses and CLIs for a moment, these models have gotten impressively smart and perceptive themselves. (While still having a jagged frontier of abilities)
frontier_modelsai_progressmodel_comparison