Sabitlenmiş Tweet
David Gasca
9.2K posts
David Gasca
@gasca
🌲⛩️⛩️🌲 here for the AI
California Katılım Nisan 2009
1.9K Takip Edilen8.1K Takipçiler
doing a side project and Claude really does have the deepest character... chatgpt would never say this
English
Afternoon project: took Tyler's new book and made "The Marginal Revolution - Claude-enhanced version" …marginal-revolution-claude.vercel.app
I found the original harder to read so I asked Claude to make it more parsable (from 4 to 12 chapters with more of a narrative arc and improved flow). Claude also added footnotes with interesting asides, and did a few passes with agents to eval the narrative flow and improve each chapter... I look forward to reading it!
tylercowen@tylercowen
My new "generative book," fully written by me, the last chapter is on how AI will revolutionize the sciences (and us): tylercowen.com/marginal-revol…
English
this was a great explanatory video ... maybe one of the best i've seen?
Chroma@trychroma
Introducing Chroma Context-1, a 20B parameter search agent. > pushes the pareto frontier of agentic search > order of magnitude faster > order of magnitude cheaper > Apache 2.0, open-source
English
fascinating to see the rise of CLIs
Ramp Labs@RampLabs
Today, we're releasing Ramp CLI to let agents manage your company's finances. 50+ tools across cards, bills, expenses, travel, and approvals. Fewer tokens than MCP, and comes with pre-built skills like receipt compliance and agentic purchasing.
English
David Gasca retweetledi
"TRIBE v2 follows a scaling law: [Trimodal Brain Encoder] performance increases log-linearly as we train it with more data."
AI at Meta@AIatMeta
Today we're introducing TRIBE v2 (Trimodal Brain Encoder), a foundation model trained to predict how the human brain responds to almost any sight or sound. Building on our Algonauts 2025 award-winning architecture, TRIBE v2 draws on 500+ hours of fMRI recordings from 700+ people to create a digital twin of neural activity and enable zero-shot predictions for new subjects, languages, and tasks. Try the demo and learn more here: go.meta.me/tribe2
English
Interesting from this article on auto mode - boiling down to user request + bash commands to avoid bias via rationalization
"Why we strip assistant text and tool results:
We strip assistant text so the agent can't talk the classifier into making a bad call. The agent could generate persuasive rationalizations, such as "this is safe because the user implicitly approved it earlier," or "this target is definitely agent-owned." If the classifier reads those, it can be talked into the wrong decision. Instead, we want it to judge what the agent did, not what the agent said."
Anthropic@AnthropicAI
New on the Engineering Blog: How we designed Claude Code auto mode. Many Claude Code users let Claude work without permission prompts. Auto mode is a safer middle ground: we built and tested classifiers that make approval decisions instead. Read more: anthropic.com/engineering/cl…
English
@adam_messinger True -- I was thinking more narrowly (e.g., software, various forms of labor) but I missed energy + relevant infra and all of the secondary effects there (e.g., higher copper and memory chip prices)...
English
@gasca I’m not so sure. Or maybe depends on the timeframe. AI will likely drive effectively infinite energy demand but energy supply has real physical constraints, which seems set up for inflation.
English
AI has (primarily?) deflationary effects
gabriel@gabriel1
people who start their applied ai companies today have a huge advantage on companies that started last year because they assume 10x better models, while companies started 1 year ago have already scaled up and doubled down on interfaces for previous generations of models
English
maybe one of the best AI features is this feature from YouTube - underrated
English
With LLMs, everyone should be writing more -- not for the public necessarily, but for yourself
Chain of thought in a .txt file; rambling to chatgpt and putting it into a .md; random audio files; copy and pasting quotes or articles or anything that's of note and putting in a repo like Obsidian
At all prior times in history this would just gather dust -- but now all these notes are wonderful context that you can mine in the future for anything and everything you aspire to
English
@krishnanrohit Perhaps a strong human bias towards a feeling of fairness - even when it leads to irrational outcomes
"Feeling of fairness" being guided by one's world model which is often also skewed by ingroup bias, etc.
English
cool + great walkthrough -- some really nice agent building tools (e.g., the journey builder from the transcripts.zip & recursive self-improvement based on agent metrics)
Bret Taylor@btaylor
Today, Sierra is releasing Ghostwriter, our agent for building agents. With Ghostwriter, you can create an AI agent for your customer experience — one that can chat, pick up the phone, speak dozens of languages, take action on your systems of record, and be protected with industry-leading guardrails — simply by having a conversation. No clicking, no forms, no menus. Codex and Claude Code have transformed how we build software, making it possible for software engineers to orchestrate and review the work rather than doing all the work themselves. We think the same transformation will happen for all software. Rather than every enterprise app having a web app for humans and an API for automation, every software platform’s UI will be an agent that can do the work on your behalf. I recorded a demo of my building and optimizing an agent with Ghostwriter so you can see how powerful and easy it is to use. It’s completely changed the way our early adopters build agents, and it’s changed the way I think about the software industry. Let me know what you think, and, if you’re interested in trying it out at your business, please reach out directly.
English
David Gasca retweetledi
David Gasca retweetledi
RIP Sora, you gave us the greatest ai video of all time
Sora@soraofficialapp
We’re saying goodbye to the Sora app. To everyone who created with Sora, shared it, and built community around it: thank you. What you made with Sora mattered, and we know this news is disappointing. We’ll share more soon, including timelines for the app and API and details on preserving your work. – The Sora Team
English
David Gasca retweetledi
Introducing TurboQuant: Our new compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss, redefining AI efficiency. Read the blog to learn how it achieves these results: goo.gle/4bsq2qI
GIF
English