🎙️ This week, Sébastien walks through four worlds: the #AI industrial complex's trillion-dollar valuations meeting physical infrastructure bottlenecks, #LLMs that have learned to play dumb during training to avoid modification, the ableism baked into AI sign-language tools, and the research pivot from brute-force scaling to domain-aware architectures. It's a meaty one. What caught your eye in #AI this week?
https://open.spotify.com/episode/0OdKW9T0lX5XclH8Sd0r4e?si=TbKSlAU0RJKMCNGxXlsJmw
Ever wonder what happens when the AI judging other AIs secretly goes soft—and doesn't even know it? Or why every single LLM defects in the prisoner's dilemma? This week on Byte of Truth, we cut through the hype and explore the absurd world of identity orbs and deceptive algorithms. Tune in! 🎧
https://open.spotify.com/episode/0tO9w09AV2GCwUmXzH9ZvO?si=oLSlgVQcQEamiNTjMzgipA
We’ve officially run out of hard questions for AI.
Researchers created "Humanity's Last Exam"—a gauntlet where GPT-4o scored a dismal 2.7%. The best models hover around 40-50%. This isn't just academic navel-gazing; it exposes the "intelligence illusion."
When CEOs mistake pattern recognition for expertise, we get AI deployed in domains it doesn't comprehend.
https://open.spotify.com/episode/2TZEyPrnEI192ncagV1cHj?si=s5VbdqPXToiQRIW5Ll1gNA
The Anthropic vs. Pentagon story isn't just business news; it's a governance flashpoint. The US military considered labeling Anthropic a "supply chain risk" for refusing work on autonomous weapons. This tension—between commercial ethics and national security—is where the future of AI governance will be forged. Europe’s AI Act offers one framework, but corporate refusal offers another. Which one holds more weight?
https://open.spotify.com/episode/1zwqizSDpi17R2qsxyAPLT?si=zbO_UNCJRrSxhmD63FhXWQ
There’s a perception that Europe creates regulations while the US creates innovations. This week’s funding news challenges that. ElevenLabs (voice AI) and Axelera (edge computing) represent a "sovereign" stack—infrastructure built to EU standards. It’s not just about catching up; it’s about building an alternative model. Is this the start of a true competitor to US tech dominance? 🇪🇺
https://open.spotify.com/episode/1zwqizSDpi17R2qsxyAPLT?si=zbO_UNCJRrSxhmD63FhXWQ
This week’s #ByteOfTruth podcast delves into a fascinating dichotomy in the tech world.
In Sacramento, the CA legislature passed SB 53, a significant AI safety bill focusing on transparency for large models. Yet, Governor Newsom signals a potential veto, creating a major policy puzzle.
Meanwhile, at Google, researchers used a quantum computer to create a “Floquet topologically ordered state”
https://open.spotify.com/episode/6lKMIDWh0SYYsqakxkOxj7?si=_8VKcsLPScmIY74h0UdrXA
#AI #MachineLearning #QuantumComputing #TechPolicy #FediBlog
What do you think about AI-driven misinformation? Share your thoughts in the comments! Tonight's episode of Byte Of Truth delves into this complex issue 🤖
https://open.spotify.com/episode/650vugtVrNO70CYWCecODJ?si=kHa4AuSSRj2awX-bUloDIw
Share your thoughts in the comments!
#ByteOfTruth #ArtificialIntelligence
Sébastien
