Conrad@innoiso
2.8K posts
Conrad@innoiso
@innoiso
Ex-Googler, MIT alum, technology enthusiast
San Francisco, CA Katılım Ocak 2020
86 Takip Edilen116 Takipçiler
As mentioned before, this is real proof that Omni hits different, which is weird that they cherry picked such bad examples beforehand
Ethan Mollick@emollick
I think people don't realize why Gemini Omni is different than other video AIs. It is fully multimodal, so it can edit video natively, too I took the famous "train " movie from 1896 & made it a bullet train, LEGO, added a time traveler, a centipede, muppets... (see reflections?)
English
Google at IO: Gemini Omni understands physics.
Gemini Omni: x.com/i/status/20576…
Google Gemini@GeminiApp
Deutsch
The Omni model is really impressive for video, but these examples are really bad.
Google Gemini@GeminiApp
Gemini Omni is here, and we’ve been seeing amazing creations all week. Here are some standouts 👇
English
Very proud that an OpenAI model disproved Erdős’s longstanding unit distance conjecture, with an elegant and intricate proof that brings sophisticated ideas from algebraic number theory to bear on geometry.
For whatever reason, mathematics has been the field most amenable to research breakthroughs with AI. I consider it lucky that it was mathematics after all - a field where experts have been willing to engage deeply with us, and with proofs generated by our models. I'm grateful for that, and don't take it for granted. Math is an artistic endeavor, and perhaps for artists, it is precisely their appreciation for art that saves them from the possibly grotesque feeling of a machine producing it.
Our goal is not to replace humans. We aim to chart a path forward where humans continue to have a significant role to play, even as we build exceptionally powerful AI. I am excited to use math as a domain to explore these paths, and @SebastienBubeck, @merettm, and I are excited to engage with the broader mathematical community to chart them together. Please reach out if you are interested!
I'm optimistic this will help us navigate how AI impacts society in domains like coding and general co-working.
OpenAI@OpenAI
Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946. For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids. An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better. This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.
English
it's important, i think, to resist the temptation to ship garbage
English
Introducing Antigravity 2.0, a new standalone desktop application that delivers fully on that original glimpse of a truly agent-optimized experience.
Rebuilt from the ground up with multi-agent teams, scheduled tasks, native voice and one-click integration with other Google products.
Learn how to get started with Antigravity 2.0 👇
English
Several data centers are now investing in their own behind the meter solutions that would make them more efficient
English
Local models are monumentally less efficient, so pushing consumers to run locally will either result in 1) less intelligence or 2) more energy expense. Unless folks have solar, this could result in more burden on the grid than centralized compute, especially because
Mark Cuban@mcuban
We should federally tax Tokens at the Provider level. Not a lot. Less than 50c per million tokens. It will accomplish 4 things (at least ) 1. It will push the big AI players to optimize tokenization, caching , routing and localization Which will 2. Reduce energy usage. Saving them in energy costs more than what they paid in tax and reducing strain created by the growth in energy consumption Which will 3. Generate maybe 10 billion dollars a year to start, but over the next ten years could grow 30x to 100x Which will 4. Create a source of funding to pay down the federal debt or deploy, in response to the things AI brings that we don’t expect or don’t like At some point the models will pass it on to customers. Of course. That’s ok. Customers will have the ability to choose between providers. Or to do everything using open source models locally. Thoughts ?
English
@joshwoodward Are you speaking again this year? If so good luck, excited for you to crush it again!
One year I'd love to be there!
English
English
@innoiso @emollick I know they're not crimes. It's an analogy.
That said, I can agree with you. It's what most reasonable people would arrive at: put the action in human terms. Grep becomes "Searching for X for Y reason", etc. I expect they're probably already making it work on working like that.
English
Codex is very good, but it is still a very "developer coded" interface for an everything app.
And it continues the somewhat annoying AI perspective that non-coders are just not as competent and need stuff hidden from them, as opposed to requiring a different form of complexity.
English
@eelheir @emollick Not quite. These aren't crimes the AI is committing. Every piece of software has to choose the level of transparency makes sense for the user. You or me might like a bit more view under the hood, but to others it causes more confusion. And so my proposal above is a translation
English
@scoopdiddy1 @emollick Yeah it might benefit from a secondary agent like the auto-review agent that just translates tool results into plain-old English
English
@innoiso @emollick giant grep awk tail chains are noise for everybody, but like error messages from an MCP? you can digest it with a Haiku and it's sooo much easier to understand. Had an insane situation the other day where an MCP wouldn't show errors because it didn't want to be """scary"""
English
They will be more confused and skeptical that the problem was solved. By contrast, Codex could set the work mode to use more understandable tools to solve the same problem. But this only matters when the user cares to see how the sausage is made
English
My take here: often, there is more than one path to achieving the goal. Codex will default to using command line tools over more user friendly tools because it means it will get the task done faster and with fewer tokens, but if the user is following the process, (1/x)
Ethan Mollick@emollick
@innoiso That is the opposite of what I am saying, though. It is not about hiding features.
English
@emollick But what's the alternative? Explaining to the user what every command line tool does? There are still some approvals that I get nervous about, even with engineering experience. But I'm not sure whether a text explanation of the tool would make me feel any more comfortable
English
@innoiso That is the opposite of what I am saying, though. It is not about hiding features.
English
@joshwoodward Does this seem in the right ballpark to you? Running the Gemma models on a Pixel 10 Pro Fold. Wasn't sure where the performance would shake out. This is a pretty simple homegrown eval, nothing special
English