Patrick Hillmann

If AI only predicts words, is it really thinking? 🤔 In this episode of The Neuron, we sit down with @evelovesolive, Founder and CEO of @logic_int, to explore energy-based models — a fundamentally different way to build AI systems that reason, plan, and evaluate multiple possibilities at once. 📺 YouTube: link.theneurondaily.com/r/EIi7C9 🎧 Spotify: link.theneurondaily.com/r/FjG5tF 🎙️ Apple Podcasts: link.theneurondaily.com/r/xHFMRl #TheNeuron #TechPodcast #EnergyBasedModels #LogicalIntelligence #LLMAlternatives

Thanks @JKFruit for sitting down with @evelovesolive. Love people waking up to the fact that we are now living in a post-LLM world.

As the world’s largest companies pour hundreds of billions of dollars into large language models, San Francisco-based Logical Intelligence is trying something different in pursuit of AI that can mimic the human brain. wired.com/story/logical-…

Logical Intelligence introduces first energy-based reasoning AI Model, and brings Yann LeCun to leadership as founding chair of their Technical Research Board The 6-month-old Silicon Valley start-up, unveiled an “energy based” model called Kona and says it is more accurate and uses less power than large language models like OpenAI’s GPT-5 and Google’s Gemini. It is also starting a funding round that targets a $1bn-$2bn valuation and has named LeCun chair of its technical research board. Most large language models answer by predicting the next token, which can sound fluent while still drifting into confident mistakes. Kona is an "energy-based reasoning model" (EBRM) that verifies and optimizes solutions by scoring against constraints, finding the lowest "energy" (most consistent) outcome. It's non-autoregressive, producing complete traces without sequential generation, reducing hallucinations. Focuses on trustworthy, math-grounded reasoning for high-stakes applications where LLMs fail, emphasizing safety, efficiency, and constraint enforcement in logic-heavy tasks like puzzles or proofs. How Kona operates Its a non-autoregressive "energy-based reasoning model" (EBRM) model, meaning it doesn't generate outputs sequentially (like LLMs do token-by-token) but instead produces complete reasoning traces simultaneously. Here's how it works step-by-step: - Input Conditioning: It takes a problem, constraints, and optional targets (e.g., a desired outcome like a proof goal or spec) as inputs. These condition the model directly, unlike LLMs which rely on probabilistic sampling. - Energy Function Scoring: Kona learns an energy function that assigns a scalar "energy" score to entire reasoning traces (partial or complete). Low energy indicates high consistency with constraints and objectives; high energy flags inconsistencies, violations, or errors. This global scoring evaluates end-to-end quality, allowing the model to assess long-horizon coherence without degrading over extended traces. - Optimization as Reasoning: Reasoning is reframed as an optimization problem. The model searches for the lowest-energy solution by minimizing the energy function, often through iterative refinement. It can revise any part of a trace mid-process, using dense feedback to localize failures (e.g., "this step violates constraint X") and guide corrections. - Continuous Latent Space: Unlike discrete token-based LLMs, Kona works in a continuous space with dense vector representations. This enables precise, gradient-based edits and efficient local refinements without regenerating entire sequences. - Output: The final low-energy trace represents a valid, constraint-satisfying solution. For example, in Sudoku, it maps allowable moves and finds a puzzle completion that minimizes energy (i.e., maximizes rule adherence). This mechanism draws from physics-inspired principles, where energy minimization finds stable states, similar to how natural systems settle into low-energy configurations. Overall, Logical Intelligence views EBMs as a path beyond LLM limitations, enabling AI that "knows" rather than guesses, with applications in verifiable, efficient reasoning. This aligns with LeCun's long-standing advocacy for objective-driven AI via energy minimization, as opposed to autoregressive prediction.

BREAKING: Logical Intelligence has signed Meta’s former Chief AI Scientist @ylecun as the Founding Chair of the Technical Research Board

Logical Intelligence brings LeCun on board as it touts AI breakthrough ft.trib.al/LpulkhV

Excited to finally be able to announce that I joined the @logic_int team as their chief strategy officer under @evelovesolive. Make sure you check our new AI system, Kona. It's wild stuff and just the tip of the iceberg for what we'll be releasing in the months ahead.

@logic_int @OpenAI @gtsoukal Verified code generation is wild 👀

Our Aleph agent, powered by @OpenAI 's GPT‑5.2, scored 668/672, 99.4% w/hyper-efficiency on @gtsoukal et al.'s PutnamBench (the hardest formal math benchmark) a critical step in natural language automated code generation — English as programming — with hallucination-free results

There has been some noise on this platform about @OpenAI's GPT 5.2 model that was off-base. GPT 5.2 can reason mathematically at a level most people haven’t clocked yet. When you force it through a deterministic proof loop, the results are startling, as evidenced by @logic_int shattering the record and scoring 99.4% on PutnamBench. Congrats to both the @OpenAI and @logic_int teams

🚀 Aleph prover just went BEAST MODE 4 math problems unsolved for 20+ years. Formal proofs in Lean 4. Less than 48 hours. Under $5k total. ✅ Binomial tail bounds conjecture (Telgarsky, 2009) ✅ Quantum gate lattice approximation (Greene & Damelin, 2015)* ✅ Erdős 124 ✅ Erdős 481 ✅ #1 on PutnamBench leaderboard The era of AI mathematics is here. Special thanks to @BorisHanin and @ylecun for helping bring this to life 🙏 And massive kudos to the @LeanFRO team — none of this is possible without the incredible foundation you've built. Aleph will be soon available to the public, stay tuned! *conditional on results from Sardari (2015), formalization pending

Congrats to @yatsiu and @animocabrands “We consider ourselves basically a levered bet on altcoins” bloomberg.com/news/articles/…

@NateMorris Why did you delete your old ESG report that you were hawking on here back in 2020? rubicon.com/blog/rubicon-i…

It's garbage day in Kentucky and thanks to Mitch McConnell, things have gotten dirty. It's time to dump career politicians and take out of the trash. DONATE: secure.winred.com/morris-for-sen…

EXCLUSIVE: @binance exec Tigran Gambaryan said that, if asked, he would be proud to serve his country fighting crypto crime under the incoming Trump administration. By @IanAllison123. trib.al/v2Nk6NZ

Nancy doesn’t miss….

@EleanorTerrett @SECGov @JohnReedStark

You know the wind is changing when former @SECGov enforcement attorney and frequent criticizer of the #crypto space @JohnReedStark is making statements like this. 👇🏼

🚨MY LATEST for @FoxBusiness: Pressure campaign secures release of crypto executive from Nigerian prison Today, @POTUS thanked Nigeria’s prez for securing @TigranGambaryan’s release, but a coordinated grassroots pressure campaign is largely responsible. foxbusiness.com/politics/press…

@SeanDuffyWI @kevinolearytv @APompliano @natbrunell @beardmars Love this @SeanDuffyWI. No better person to host it. Agree with folks on here that @EleanorTerrett has earned the respect of the crypto business community, but I’m sure you guys have plans for later shows. It’s just the first show people!

Fox Business has a new Crypto show and I am hosting it with all the best experts! Make sure you check it out tonight at 8:00 PM ET on Fox Business. Kevin O'Leary - @kevinolearytv Anthony Pompliano - @APompliano Natalie Brunell - @natbrunell Marshall Beard - @beardmars Hester Peirce - @HesterPeirce Dan Gallagher - @DanGallagherDC Brad Myers - @JBradMyers

Release Tigran Gambaryan!

I know what country he’s a citizen of, where he lives, his “nationality,” and a hobby he has that would dramatically narrow down his identity for someone inclined to investigate. But why does it matter? If he wants his anonymity, he should have it.

There are three people in the USG that I am aware of who know the actual identity of Satoshi. I am very close with one of those individuals. While that person will never tell me who it is, I do know for a fact that it is not Peter Todd...nor any other North American for that matter politico.eu/article/peter-…

@WFascination @allmiin Not Asian

@PRHillmann @allmiin Enough to know if he's Japanese?