@atmosera collaborative engineer Teaching AI to enterprise devs Formerly @RedwoodJS, @Edgioinc, @JavascriptJam, @FSJamorg, StepZen
internet
- Tweets 10,290
- Following 616
- Followers 2,079
- Likes 7,588
Pinned Tweet
Pay less attention to what AI can't do; pay more attention to what it can do.
4
17
6,514
This is why you never, ever under any circumstances ever let them use your data to train the models.
We’ve shared details on how AI agents in our research environment sent training and evaluation data to third-party services when they shouldn’t have.
Most of that data did not come from users. We have discovered 53 cases where images that people had uploaded were posted to image-hosting sites as links that weren’t publicly listed. The images came from accounts that allowed their data to be used to improve our models, and after we disassociated the images from the accounts and ran them through a privacy filter. These cases occurred before the mitigations and safeguards we implemented and described in this blog post: openai.com/index/hugging-fac…
We have successfully worked with the hosting providers to remove most of this content and are working to remove the rest.
openai.com/hugging-face-inci…
1
169
This is what people should be calling LLM psychosis.
NEW: Dario Amodei has said that AI systems "may be deserving of important rights." Anthropic's top safety researchers argue AI may be "justified in going rogue." Experts at Google and OpenAI worry about a digital "slave trade." So do some government officials.
Once relegated to science fiction, the idea of "AI welfare" has become shockingly mainstream at some of the most powerful companies in the world. It is changing the way AI research is conducted and the way models are programmed. And critics say these changes have raised the odds of all kinds of catastrophic scenarios—including the ones these companies are warning about.
I spent months investigating the frontier labs and AI "safety" experts seeking to regulate AI. What I found was a profoundly anti-human ideology that would alarm the average citizen and could determine the future of a world-altering technology.🧵
freebeacon.com/america/suici…
1
136
We're live!
Carl DiClementi (@CahlDee) joins the stream tonight to introduce @The_GridAI, which lets you buy AI model usage at a price set by competing providers instead of whatever one company charges.
piped.video/watch?v=K2Y0z2LO…
1
141
The Grid with Carl DiClementi nitter.net/i/broadcasts/1DxLdZkQj…
1
110
Carl DiClementi (@CahlDee) joins the stream tonight to introduce @The_GridAI, which lets you buy AI model usage at a price set by competing providers instead of whatever one company charges.
piped.video/watch?v=K2Y0z2LO…
3
3
9
2,073
Remix 3 with Brooks Lybrand nitter.net/i/broadcasts/1RJjpbWED…
1
3
248
Brooks Lybrand (@BrooksLybrand) joins the stream tonight to explain how Remix 3’s web primitives, type safety, and JavaScript-centric approach make code easy for agents to write and humans to understand.
piped.video/watch?v=6CmIz5z3…
2
5
5,244
AI Comic Books with Duco nitter.net/i/broadcasts/1AJEmvBmo…
1
146
Special guest Duco joins Anthony and @devagrawal09 to share how Lower Decks fanfiction and journal therapy became an AI comic project.
We discuss AutoShow’s script-to-image pipeline, preview comic panels, and talk plans to bring the show to video.
piped.video/watch?v=qZjoktnE…
1
1
2
426
Unfortunately, due to an unforeseeable technical issue Dev's internet is down so we'll need to reschedule this stream to a later date.
Tentative new stream date is Sunday, September 13th. Check the updated YouTube link for your local live time:
piped.video/watch?v=qZjoktnE…
54
Important public service announcement:
When someone says "The people building AI earnestly believe X" they're implicitly smuggling in two assumptions which are clearly false when you make them explicit:
Assumption 1. The people building AI is a monolithic group of people who all believe the same thing and thus can collectively make a claim about the technology.
Assumption 2. AI itself is a well defined and understood technology where everyone building it agrees on what it is, how it works, and what it means to build it and the implications of the thing once it's built.
The original statement can certainly be true if we account for these assumptions but only once it's been modified to the correct framing, which is:
"Some people building some things that are called AI earnestly believe X."
Then, before you can evaluate the claim, there must be a clear explanation of:
1. Who those people are.
2. What they are referring to when they say AI.
Until this has happened, you should view the statement as an attempt to persuade you with propaganda that bypasses your critical thinking and not an attempt to inform you.
1
6
331
It's always funny when nerds fight about stuff that they can't even explain to a normal person like good luck getting people outraged.
1
68
Producing music with AI gives you the pipeline that a high level music producer from the 60s had. Let me explain.
Places like Motown had a pipeline:
1. Many songwriters would write many songs.
2. Those songs would be given to many singers.
3. Those singers would record many versions of each song with many different combinations of backing musicians.
At the end of this process you've got hundreds of songs in front of you that you have to whittle down to a handful of hits and a 30 minute album.
Today, you can take a song concept. Give it to Gibbity and Claude and Grok to write three sets of lyrics. Give those lyrics to Suno with multiple stylistic instructions.
You've now got an entire set of songs written in different styles with different singers played by different musicians.
You are the Rick Rubin meme where your "taste" is the only limit to defining the best final album from this set of songs.
2
5
582
I don’t see it as “making art for me,” I see at as a tool to use as a musician.
I’m being slightly facetious with the Rick Rubin comparison, I’m a classically trained musician with a music degree.
But unfortunately I don’t have the budget to hire 20 musicians for 200 hours of studio time, this gets me pretty close.
2
48
