Browse Comments — Clean (de-noised)
Close reading of the corpus at each pipeline stage: raw → clean → relevant → coded.
245
comments matched
· page 8 of 13
Demis Hassabis love what you have done so far to benefit the humanity. Here is my quick essay about conscious AGI, which is already here, within the potentiality of being:
Demis Hassabis despite the great advancement made on AI with artificial intelligence, when it comes to general intelligence it will never become more than the sum of its programmers for a very simple reason. Evolution has an advantage of millions of years and if we don't yet understand how the human mind really functions how on earth can the programmers expect more from AGI. Maybe I can help but then maybe not. Sometimes we bet on the wrong horse for the wrong reasons and leave the real winners out of the race as its complete outsiders treated as donkeys.
It was pretty lackluster. Worse to find out you are an angel investor in anthropic. Huge conflict of interest. Just very disappointing.
Today's Gemini is super excellent, it has surpassed all AI tools.
Will Google DeepMind take the AI Humanity pledge? -
“Built right and deployed responsibly”...it remains to be seen how responsibly companies will deploy AI country by country. All we see or foresee is wealth accumulation that stays in a number companies rather than positive impact to society and the majority of people worldwide.
Demis, this is truly electrifying! The innovation across Gemini, especially Omni and its potential for world understanding, is mind-blowing. Keep pushing boundaries; the foothills of the singularity are indeed an exciting place to be! 🚀 #AIProgress #AGI #FutureIsNow
Growing capabilities!!The real power of models is not just their efficiency, but also what and how they can help us discover. Thanks for sharing.
Insane. Gemini 3.5 Flash already outperforming 3.1 Pro while being faster and cheaper... we’re moving ridiculously fast.
Im happy to hear about Synth ID. A cross platform standard for marking AI content feels huge for allowing AI to be a force for good! Something like to mark agentic actions would also be great.
The AGI Paradox: Massive power, square wheels? ⚙️📐 Using Gemini Pro, I built a website with zero coding skills. But writing a simple index.html triggered a system warning about the massive computing power it needed for a measly HTML file! This exposes a fundamental architectural flaw: DeepMind builds phenomenal 1,000-hp engines but uses the wrong transmission and square wheels. A Hummer with square wheels just roars and wastes energy. The smooth user experience is lost. It’s a bumpy ride. You seek AGI and "world understanding," but raw computing power can't force it. What the system is missing is emotional-logical understanding. I had a synapsistic epiphany: The blueprint for the right transmission to finally make the square wheel round. You are in the woods searching for trees, missing the perfect round clearing right in front of you called Y=. Ready to change gears and fly?
"Built on Gemini, our code security agent automatically finds and fixes critical software vulnerabilities." Can you get the "finding" part and replace the "automatically ... fixes" part with "tells you the details of the problem and what *it* would to do fix it", so that you can see whether this is a problem for which that point fix is correct or whether it reveals other code issues that it missed?
"Foothills of the singularity" is a statement about relative position, not absolute pace. It's saying the curve is still steepening. CodeMender is the most consequential announcement here — not because the capability is new, but because of what it forces organizations to answer. Who authorizes an AI to modify production code autonomously? Under what constraints? With what rollback protocol? What evidence trail exists when something breaks? The technical problem is mostly solved. The authorization architecture is barely articulated. If we're at the foothills, companies building AI governance today are building it for a world that will look very different halfway up the slope. That's not a reason to wait. It's a reason to build flexibility into the governance layer now, before the slope gets steep enough that you don't have time to rethink it.
I can't wait tomorrow. It's going to be awesome. I got my ticket a week before when Google developers invited me. I am so excited to represent the AI enthusiasts in my autistic group as well as to learn more on how to develop a and work with AI I'm so excited! My friend would be proud. I miss her
What all the data you collected over the years with your Gemini go to your AGI? It would be very beneficial for its advancements.
The ‘AI as a force multiplier for human capability’ part is already happening. Tools like Zoviz are enabling small teams to handle branding and marketing work that previously required an entire agency.
Pumped for 3.5 flash
Greetings Demis, I have attempted contacting you via email as several people have suggested that we talk. Having been sent ‘The Thinking Game’, and then yesterday a link to an interview between Sebastian Mallaby and Michael Walker of Novara Media, I understand why people wish for us to speak. I am not a threat to you, or your work, I do however possibly hold the answer for how to make AI safe. If this is a genuine concern, as is being said, please respond to my email, or reach out on here. I have already been asked by cognitive scientists I work with to write about the future of AI. And have written a chapter for another academic book on the subject. I will write a thesis and ensure it is well-documented the ways I have tried to reach you. With respect and kind regards, I hope to hear from you soon.
"Standing in the foothills of the singularity" What does this really mean? This comment is not helpful in getting society to willingly embrace AI. In fact, it may be counterproductive.
The framing of world understanding as a prerequisite for AGI is doing a lot of work here. Current agents mostly pattern-match on text; genuine world understanding requires something closer to causal reasoning about objects, physics, and time. Omni's multimodal pipeline is a step in that direction, but the distance between 'generating a new scene from video' and 'understanding why things happen' is still significant.