Haider.
@haider1
together, we build an intelligent future.
3.8K Following 66.3K Followers
Sam Altman says humans may not be smart enough to unlock biology's deepest secrets, but we don't have to be
An intelligence a million times smarter than us could crack what we can't, then explain it in a way we can understand
"we don't need to discover it, just understand it once someone does"
Show more
so far, these are the likely confirmed updates ahead of google i/o:
new ui and gemini 3.1 flash lite will be included in the gemini app
gemini omni, likely focused on video
gemini 3.2 or 3.5 flash and pro
new gemini "spark" agent
the "avatar" option linked to gemini omni
Show more
so far, these are the likely confirmed updates ahead of google i/o:
new ui and gemini 3.1 flash lite will be included in the gemini app
gemini omni, likely focused on video
gemini 3.2 or 3.5 flash and pro
new gemini "spark" agent
the "avatar" option linked to gemini omni
Show more
one important thing to remember:
openai is scaling compute much faster than anthropic, and adding many of the computer-use features that made claude attractive to enterprises
Enterprise customers don't care if claude is more natural to talk to
they go where the safest bet is
Show more
Sam Altman says humans may not be smart enough to unlock biology's deepest secrets, but we don't have to be
An intelligence a million times smarter than us could crack what we can't, then explain it in a way we can understand
"we don't need to discover it, just understand it once someone does"
Show more
one important thing to remember:
openai is scaling compute much faster than anthropic, and adding many of the computer-use features that made claude attractive to enterprises
Enterprise customers don't care if claude is more natural to talk to
they go where the safest bet is
Show more
Creator of C++, Bjarne Stroustrup:
AI-generated code isn't ready — it generates more bugs, more bloat, more security holes, and is nearly impossible to validate
"senior developers are already retiring rather than deal with it"
The problem is that even a small prompt change can shift the entire codebase in unpredictable ways
Show more
Creator of C++, Bjarne Stroustrup:
AI-generated code isn't ready — it generates more bugs, more bloat, more security holes, and is nearly impossible to validate
"senior developers are already retiring rather than deal with it"
The problem is that even a small prompt change can shift the entire codebase in unpredictable ways
Show more
google is in a different position
openai and anthropic are pushing hard into coding because that’s where the near-term money is
but google already makes huge money from cloud, owns TPUs, invests in anthropic, and still releases research that other labs use
so it's more do-or-die for oai and anthropic
Show more
google is in a different position
openai and anthropic are pushing hard into coding because that’s where the near-term money is
but google already makes huge money from cloud, owns TPUs, invests in anthropic, and still releases research that other labs use
so it's more do-or-die for oai and anthropic
Show more
i try not to overhype things too quickly
but for the last 2 days, it really looks like openai has been testing a new model version inside chatgpt, both for text and image
i've repeatedly gotten stronger responses from different a/b test versions, and one is lightning fast
Show more
google will probably announce a new model in the next 2-3 days
i'm hoping they jump straight to gemini 4.0 instead of doing a 3.2 or 3.5 release
gemma 4 is already out, and gemini 3.1 came pretty quickly after 3.0, so google's usual version pattern doesn't predictable anymore
Show more
grok 4.3 has become one of my most-used models
last night, i used the same prompt to write automation scripts with opus 4.7 and grok, and grok reached a better final product
while opus ran out of tokens before i could add QoL features
hope xAI releases a local open-source, like gemma or gpt-oss
Show more
really cool benchmark for long-horizon test-time adaptation
gpt-5.5 in codex leads on FutureSim, where agents interact with a chronological replay of real-world news and are tasked with predicting future events
on some Polymarket questions, gpt-5.5 even moved ahead of the human market aggregate
interestingly, gemini 3.1 and opus 4.7 are missing
Show more
Continual learning is bottlenecked by realistic evaluations
Introducing FutureSim, which replays real-world events in the temporal order they occurred
We benchmark frontier agents at updating predictions about how our world evolves, in native harnesses like Codex, Claude Code
Show more
Yann LeCun says LLMs are strongest in domains where language itself is the substrate of reasoning, like math and code
They can solve problems, prove theorems, and write programs — but they are not creative mathematicians, software architects, or computer scientists
"their role is to help humans build"
Show more
