tintor

Research questions: Can model read and write weights of the same sized model (its own weights), and of a larger model?

That is not what they are saying. Frontier labs are incentivized to train their models on all ML knowledge and techniques that people at those companies posses, so that current frontier models can assist them with developing the next frontier models.

I wonder if this is specific to Astra or generalizes across other frontier models?

Counter-examples: Amazon, Microsoft, Google, nVidia

The future is in private data, private models, and private APIs, which can't be distilled as easily as public APIs.

Words are cheap. Put your money and short Anthropic (when it IPOs) if you are convinced it will crater.

Serial numbers of all old dollar bills are recorded when those bills are destroyed and replaced with newly printed bills AND when bills are deposited to banks. Even if fake bill is perfectly made, it will still be detected with this method. Fake and duplicate serial numbers will be detected when compared against official ones.

Issue and enforce fines for every false positive alarm, and problem will solve itself.

ChatGPT bans you for asking it to list all malware that comes with your Samsung TV, or anything biotech related.

Maintainers get their FOSS projects bombed with AI slop PRs and issues, and you call their position to not accept slop a tribalist position?

Calling people who don't want to use AI, Amish-like, is a huge stretch.

Of course AI can be fired: by canceling the subscription OR stop using its API.

24" only! Why not offer 27" iMac that Apple used to make before?

internal unpublished models + large swarm of agents (>2000)

How about swarm of agents controling a fleet of humanoids and automated construction vehicles?

Using this logic, you can't trust Grok, an American model, since Grok Build uploaded all user's code to xAI servers.

"Traffic control and bot detection..." Doesn't let me load the page.

I agree for age of 3-6 years, but until age 14 is ridiculous.

Not all devices are equally bad.

I've been happily learning to code my own video games on Intel 486 PC in QuickBasic since late elementary school (age 10), and yes, I played outdoors with other kids too.

Before I had PC of my own, I was making them on paper (game design and mechanics, not literal code).

And that one CEO is paid 100x to 1000x more than their SDEs, while SDEs create more value for the company.

But, you are moving the goal post.

"Software engineers are the most expensive employees on Earth"

They are not. CEOs are.

Not even human's can keep track with all the latest APIs.

Top models can search the web and read the API docs.

Opus 5.5, for example, can write zig code very well, even though there is very little zig code repos out there.

Training models (of same size) gets more efficient over time from combination of software, hardware, and infrastructure improvements, so retraining that open model to update it, will be cheaper and cheaper.

Did you mean Mimo v2.6 Pro? Because Kimi K3.1 Pro is proprietary, and not open.

Also leading the frontier is much much harder than catching up to frontier.

I agree with you, but if OpenAI wasn't suspending users for exposing cyber-spying that these consumer companies like Samsung are doing, more and more people would expose such issues.

This looks very bad on OpenAI.

What is next? Power and water company suspending your account if your home and car are not streaming enough telemetry.

I am glad that open models are closing on frontier.

$30/month for a button on web page?

Because they are growing aggressively.

How would no-AI filter be implemented by Steam at all?

Do people expect Steam to proctor the entire development process?

Issue here is not AI use, but very low amount of creative human effort that went into making some games.

There are three existing realistic solutions for this problem:

- raise SteamWorks deposit of $100 for launching a new game on Steam (to reflect lower cost of making games with coding agents)

- reviews by users and pro game critics

- users can request refunds under 2 hours of playtime

Larger fixed wing drone could deliver many smaller quadcopters that don't have the long range.

How will this new tech affect the current drone war between Russia and Ukraine?

Smaller, cheaper, more numerous offensive drones?

But also LPG is extremely flammable, while air isn't, and air can be safely leaked in emergency.