Browse Comments — Raw (as collected)

This is the same discussion as the one people had when smartphones took of, when the internet became popular, etc... And I see no reason for it not to end in the exact same way as the previous did.

A powerful reminder that AI’s real potential lies in amplifying human creativity, research, and innovation — not just automation.

Operating at the frontier requires a massive infrastructure pivot. Rising capital expenditure on data centers, energy grids, and watermarking frameworks proves that software capability is bound by physical world resource limits.

Francesco Nicoli but harder to use AI in a fast paced simulation in which you have to think in real time and the grade is on the performance in the simulation. So the assesment is not to deliver a game/simulation, is the performance in the game. In a sense orals are a game of this kind. It has rules, victory/losing conditions, score, entry and exit points.

Can we get some projects on Scratch?

Demis Hassabis you are such an inspiration, one of the few driving AI to benefit human progress. Many tech CEOs and leaders should learn from you.

Reading AI announcements in 2026 feels like watching humanity speedrun the tech tree. ‘The model now understands video, audio, code, science, agents, and reality itself.’ At this rate, next year’s keynote will just be: ‘We asked Gemini to present I/O. It felt our version was inefficient.’ But beneath the hype, there’s something genuinely historic happening: the industry is moving from chatbots that respond to systems that can reason, act, create, verify, and collaborate. That’s a much bigger transition than most people realize. Also, calling it the ‘foothills of the singularity’ is probably the most Silicon Valley way possible to say: ‘Things are about to get very weird, very fast.

The shift from isolated generative outputs to autonomous agentic execution presented at I/O alters the structural economics of professional knowledge work. At Lex Experience, our architecture for elite legal BD relies on sustained multi-step reasoning. We assess that Gemini 3.5 Flash forces a structural recalculation through (i) a 1,048,576-token context window directly optimized for parallel agentic execution loops; (ii) documented outperformance on the MCP Atlas benchmark against the 3.1 Pro baseline; and (iii) the explicit integration of SynthID across Omni media, which enforces the chain of custody demanded by institutional risk management. While AGI represents the long-horizon trajectory, the immediate friction point remains governance: deploying these frameworks safely requires that validation protocols scale exactly in tandem with 3.5 Flash’s execution velocity.

Rubén Domínguez Ibar I am the Moat

Food for thought.

Paolo Spada yes sure. I was thinking more in terms of research. I recently used codex to build a custom simulation software that implements a formal model of European integration I had developed over the past 10 years or so. Without codex I probably would have never finished it, nor build an entire dedicated simulation software.

Dr. Dean Fido CPsychol AI review is a large amount of work, often outside personal expertise at least in politics that is heterogenous as a field, often operating from pilot processes (we have a new AI declaration the students sign that is the basis of the review this year) and if it goes rampant, I had 26 cases in a class last year, it is a month of work that is not properly workloaded. The advantage of our new AI declaration is that it allows for class level customisation. So in one of my classes I piloted a -30 points flat punishmrnt for hallucination they can challenge with an office hour discussion with me that would determine if it was a honest mistake instead. Got a few, nobody challenged them. I did not involve AI officers. But agentic AI can mechanically cross-check references, particularly if you load all the sources, so this way of catching AI misuse is disappearing. An agent writes the essay or corrects it adfing references, one downloads the pdfs, one does the cross-check, ranks probability of hallucinations, and then the student does a manual cross-check. Maybe a couple of hours of work, perfect referencing, zero reading and learning (apart for a bit of AI agent setup following youtube videos).

What stood out to me from Google I/O 2026 is how Gemini is rapidly evolving from a conversational AI model into a broader multimodal and agentic ecosystem platform. I think this could have major implications for India’s enterprise, developer, and AI startup landscape. I have explored this in my analysis for PC Quest https://www.pcquest.com/news/google-io-2026-gemini-conversational-ai-india-analysis-11853177

Paolo Spada 'Given we do not ask for replication packages we have very reasonable looking stats output that we doubt are real. ' nailed it! Nice read and thanks!

Since the floor has been raised, maybe it makes sense to start asking for replication packages from undergraduate dissertations.

Absurd. And how statistically significant were the results?

Flash 3.5 is my go to model for now 🙌💛 The cost savings vs opus is a no brainer

I just started my journey in AI automation engineering 🫠 and I think this resource will be useful. Thank you for sharing 😊 Abhishek Veeramalla

How amazing you are !

Bold and timely, Edgar Perez Global Speaker!Jensen’s “forget about China” cuts through the noise, thank you!