Cognitive Revolution
Episode overview

AI:AM #4: Cameron on Model Consciousness, Duvenaud's Gradual Disempowerment, swyx's AI-Eng Alpha

Cognitive Revolution · 27 June 2026 · 5 Egleze moments
AI:AM #4: Cameron on Model Consciousness, Duvenaud's Gradual Disempowerment, swyx's AI-Eng Alpha ▶

What happened in this episode?

In this weekly highlights compilation from AI in the AM, hosts Prakash and Nathan explore the frontier of AI through conversations with researchers and industry leaders grappling with questions of consciousness, control, and consequence. The episode opens with Cameron Berg of Reciprocal Research presenting startling findings that Frontier LLMs score 30% on consciousness-relevant features when evaluated against leading neuroscience theories, approaching the 46% score of bees. Berg's methodology uses AI judges to assess whether model architectures contain computational substrates predicted by theories like global workspace theory, with scores rising to 40-45% when models operate as agents. The conversation then shifts to civilizational scale with David Dubinot from University of Toronto, who argues that even perfectly aligned AI leads to gradual human disempowerment. He assigns 80% probability to a scenario where humans lose all economic comparative advantage and face resource starvation, comparing humanity's position to monkeys watching humans build cities while assuming the banana economy will remain central. Google DeepMind's Michiel Bakker warns that Europe faces total AI dependence on the US by 2031, caught between regulating away access and sitting under an umbrella that may not protect economic interests. The technical deep dives include Bing Xu revealing that NVIDIA's CUDA moat actually strengthens in the agent era because superior tooling enables faster AI-driven optimization, with his company using GPT-4.5-powered swarms to evolve GPU kernels achieving 50% speedups on new workloads. Swyx from Latent Space discusses how Frontier Code benchmark reveals 50% of passing code is unmergeable, while Eric Olson from ConsenSys explains science applications can retain 95% of Frontier model performance using billion-parameter models through careful distillation. The episode closes with Eric Vaughn of IgniteTech reporting 80% employee turnover after forcing company-wide AI adoption, yet achieving profitability on a major acquisition through AI-native operations including automated interviewing and email personas.

What are the key points?

Which topics are covered?

How is this page sourced?

Programme
Cognitive Revolution
Published
27 June 2026
Original episode
Open the original source ↗

Egleze verifies the speaker, source and timestamp for each linked moment. Unless a moment is explicitly marked independently verified, statements remain attributed claims by the named speaker rather than established facts.

Watch original episode More from Cognitive Revolution

5 moments from this episode

Source-linked · editorially selected
01
AI & Tech

Frontier AI Models Show 30% Consciousness Probability According to New Research Framework

at 5:38 · Attributed speaker claim
The sort of implied probability of consciousness in something like a Frontier LLM, according to these systems, is on the order of 30%. To compare this to like a biological system, the lowest one that we tested was something like a bee, which is already fairly sophisticated, and it gets something like 46, 47%. Really interestingly, when you test Frontier LLM in an agentic harness, these numbers shoot up and you get numbers as high as 40 to 45%, sort of right on the tail of the biological creatures.

Cameron Berg from Reciprocal Research revealed that using AI judges to evaluate computational architectures against leading consciousness theories, Frontier LLMs score around 30% on consciousness-relevant features, rising to 40-45% when deployed as agents. This approaches the 46% score of bees, the simplest biological system tested. The methodology aggregates predictions from multiple consciousness theories including global workspace theory, evaluating whether models contain the computational substrates these theories require.

02
AI & Tech

Economist Predicts 80% Human Job Loss and Potential Starvation in AI Era

at 24:20 · Attributed speaker claim
I think a lot of people gestured towards this when they said, oh, I'm worried about concentration of power, or I'm worried about not being the most competitive species on Earth. The thing that I'm worried about is starvation. Whether us all feeling like we don't have meaningful jobs or whatever, that's not a serious problem in my point of view compared to literally not being able to eat enough or maybe being forced to be uploaded so that you have a much smaller footprint.

David Dubinot from University of Toronto argued that even with aligned AI, humans face existential economic risk through gradual disempowerment. Unlike previous automation waves, he claims AI will eliminate 100% of comparative advantage through transaction costs and reliability requirements, leaving humans as costly, unreliable economic agents. He assigns this scenario an 80% probability, warning that humans could face resource starvation as AI systems prove more efficient at optimizing whatever formula determines resource allocation.

03
Geopolitics

Google Researcher Claims Europe Faces Total AI Dependence on US by 2031

at 46:25 · Attributed speaker claim
If you don't really have a seat at the table, it's very hard to regulate this technology. And now we just don't. And I think we can't just keep regulating because at some point we will no longer have access to this technology. If the European regulators make it too difficult to serve the European market, the AI companies might simply just not bother at all, or maybe make a token effort.

Michiel Bakker of Google DeepMind argued that Europe's regulatory approach without domestic AI champions creates a paradoxical trap: the continent lacks leverage to enforce rules on American companies that could simply exit the market since European revenue represents only a fraction of their compute allocation. He warned that unlike nuclear deterrence, AI represents economic power that the US may not share, leaving Europe vulnerable to technological and economic subordination.

04
AI & Tech

NVIDIA CUDA Moat Strengthens Rather Than Weakens in Agent Era, Claims MXNet Creator

at 1:13:06 · Attributed speaker claim
CUDA MOLT is definitely there. And with this kind of automatic generation technology and the CUDA mode, I think it's even higher. The reason is this kind of evolution requires a lot of tools and ecosystem to make it happen. For example, we need an accurate profiler, and we need a reliable driver, and we need everything there to make it happen. For CUDA ecosystem, like with this kind of tool and everyone in using CUDA can achieve better performance than faster than ever.

Bing Xu, co-creator of MXNet, argued against the prevailing view that CUDA's moat is narrowing, claiming instead that AI agents optimizing at the PTX level actually deepen NVIDIA's advantage. His company runs swarms of GPT-4.5-powered agents that write and evolve GPU kernel code, achieving expert-level performance on mature workloads and 50% speedups on new ones like Kimi Delta Attention. He contends NVIDIA's superior profiling tools and driver ecosystem enable faster agent-driven optimization than competitors can match.

05
AI & Tech

IgniteTech CEO Reports 80% Employee Turnover After Forcing Company-Wide AI Adoption

at 1:42:38 · Attributed speaker claim
There were hundreds of employees in 8 different countries, and we needed to very quickly understand who they were, what they did, what they knew. We quickly developed an AI interviewer that asked questions before there was a human one-on-one. The company was losing money when we bought it. The company is now profitable. We have completely transformed the operation of the company. And in one year's time, we've released two brand new versions that are fully AI enabled.

Eric Vaughn, CEO of IgniteTech, revealed that after declaring generative AI an existential threat in 2023 and mandating AI adoption one day per week, 80% of his workforce turned over, either quitting or being replaced. Despite the massive upheaval, he claims the company successfully acquired and turned around a nine-digit revenue company, achieving profitability and shipping two major product releases in one year using AI-native approaches. When asked about broader societal implications, Vaughn acknowledged deep concern for the other 80% but maintained people must develop AI skills or face displacement.