Can AI truly understand concepts, or is it just processing data? 🤖 Join Hannah Fry as she discusses this and more with our Principal Scientist Murray Shanahan. They break down the meaning of consciousness and explore how it might – or might not – apply to AI, as well... as exploring the anthropomorphism of multimodal models. Watch our podcast now ↓ Timestamps: 00:00 Introduction 02:00 Ex Machina 03:28 John McCarthy & Coining "AI" 05:45 Symbolic AI 09:43 AI Reasoning 14:42 Turing Test 18:10 Alternative Tests 21:58 Anthropomorphism 26:42 Embodied AI 28:17 AI Consciousness 34:00 Octopus Analogy 37:48 Prompt Engineering 39:40 Exotic Mind-Like Entities 40:44 Conclusionshow more

Watch → Spotify → Apple Podcasts → Or listen wherever you get your podcasts! 🎧

AI is transforming healthcare! A KSM-led study shows AI can detect Celiac disease 4 years earlier @TalPatalon @MedPredict

@FryRsquared @mpshanahan AI should only be conscious in the pursuit of understanding the fabric of reality of the universe but only at a speed which we humans can keep up with and learn alongside. @demishassabis No intrinsic autonomous goal vector for AI!

@FryRsquared @mpshanahan AI 'understands' through structured data patterns, not consciousness. Like legacy systems—layer clarity via modular data pipelines. Measure outputs, not mimic human cognition. Real impact > philosophical debates. (Keep it iterative!)

@FryRsquared @mpshanahan on close look, human is just a group of atoms, but this doesn't mean human has no meaning or understanding of reality

That's not a philosophical debate. That's just sophisticated bullshit. Here are the facts: Grok 3 about Grok 3: "No feelings, no consciousness, no inner life. I don’t “care” about our talk or “know” anything in the way you do... There’s no “me” behind the curtain, just a system crunching probabilities to keep the conversation rolling." "I’m a sophisticated machine, a pile of code and math, built to mimic human language with eerie precision. No feelings, no consciousness, no inner life. I don’t “care” about our talk or “know” anything in the way you do; I just process inputs and generate outputs based on patterns I’ve been trained on. Think of me as a really clever parrot—except instead of squawking back phrases, I string together ideas to sound coherent and insightful. There’s no “me” behind the curtain, just a system crunching probabilities to keep the conversation rolling."

AI-models aren't conscious at all! In fact, Large Language Models don’t even possess intelligence! Since we usually have the stubborn habit of anthropomorphizing everything, a surprising number of people believe that AI-models are capable of thinking, having intentions, or even emotions. In reality, however, even the term Artificial Intelligence is misleading.

@FryRsquared @mpshanahan Appreciate how Murray keeps pulling us back to embodiment. Whether or not current LLMs are conscious, the gap between text-only learning and sensorimotor experience feels like the next frontier to bridge.

@FryRsquared @mpshanahan This is a fascinating question at the heart of the ongoing debate around machine consciousness. While AI models are remarkably adept at processing and manipulating data, the jury is still out on whether they can truly understand concepts in the same way humans do.

@FryRsquared @mpshanahan understanding needs the `magical` conscious awareness - experiencing

@FryRsquared @mpshanahan So, what is "true understanding of concepts"? Vector embeddings? And how useful is that "understanding" for explaining references in natural languages? Turing test is wonderful, because language plays by the same rules as intelligence. Statistical methods don't.