Matthew

tech guy, passion for learning, innovation, and disruption; slight problem with authority, what can I say.

[Matthew @MBall_Atlanta]

https://t.co/E4fIBtQe6j

[Matthew @MBall_Atlanta]

@finkd So you agree with safety and alignment you just don’t need the same bailout the others require to remain solvent. 😀

[Matthew @MBall_Atlanta]

@CompleteSkeptic 👀🔥🤞

[David Sacks @DavidSacks]

Dario has written that we need to “pace the frontier,” and Sam has agreed. People may be surprised by my response: go ahead.

You guys are the frontier. By any reasonable metric — market share, revenue growth, model capability — the two of you have a duopoly on frontier intelligence. You’ve also claimed the lead is widening because of recursive self-improvement.

I don’t see what you see in the lab. If the unreleased models are scary enough that you think you should slow down, I support your decision to be responsible.

But stop pretending you need anyone else’s permission. Stop pretending antitrust law has to be suspended so you can form a cartel. Stop pretending you need a regulatory approval process that supersedes product liability. Stop pretending METR is independent when it is intertwined with Anthropic’s investors and staff. Stop pretending you need those same evaluators to police competitors who aren’t even at the frontier.

Most of all, stop pretending the motivation to slow down is purely altruistic. You face massive product-liability exposure if your products enable a truly damaging cyberattack. The market already punishes models that behave in unpredictable or unauthorized ways. After the Hugging Face episode, it is simply good business for OpenAI and Anthropic to trade some raw power for reliability and predictability. Call it alignment if you want. It is also just giving customers what they want.

Pacing the frontier would also create breathing room for a more intelligent conversation about regulation than Bernie Sanders’ “shut it all down.” China is very unlikely to join a global agreement, as you know, and that has to be taken into account as well.

So go ahead and pace the frontier. You are the ones setting it. The easiest way not to build superintelligence is for you to agree not to build it. Demanding your preferred regulatory framework as the price of that will look like blackmail of the public and the political system. So just do it.

If you do, you’ll buy goodwill for the next conversation. If you don’t, we’ll know this was just another bid for regulatory capture — or an election-season psyop.

[Matthew @MBall_Atlanta]

@kirawontmiss I don’t know exactly how cybercab program works but why can’t uber drivers who already often buy tesla just buy a cybercab, let it work all day for them while they have the possibility to hold another job…improving their quality of life. Not a possibility?

[Matthew @MBall_Atlanta]

@Zephyr_hg Articles used to suggest quality research, now they are fancy wrapper for useless context and a link to a paid product.

[Matthew @MBall_Atlanta]

@rohanpaul_ai Maybe this is correct for some companies and not for others. I would defer to the company experts of AI engineers architects and other roles to decide.

This reads to me like a plea to reduce the burden on their audit firm 🤷🏻‍♂️

[Matthew @MBall_Atlanta]

To each is own, but this comment doesn’t fly in many company settings if you are responsible for AI strategy. Let me try this with our chief legal and security officers…

But yet 1000 skills and agents running around, in my example…let’s not govern we don’t yet understand…

Again, governance evolves just like the models and the level of autonomy, etc. you can very well start with governance 101: do you have a policies, an inventory and know the risk classification at the very least.

[Matthew @MBall_Atlanta]

It’s more art than science but essentially after basic adoption and the skills and agents are showing signs of momentum you should already be knee deep in your approach even if vaporware to start, with your governance counterparties. It’s more in parallel…

All I am warning against is if governance isn’t evolving in their thought process and understanding by the time you have 1000 agents and skills running around you are in for some real friction and challenge.

[Matthew @MBall_Atlanta]

@mardehaym Makes sense and companies are totally missing it. I’ll research your other materials thanks for pointing that out

[Matthew @MBall_Atlanta]

Agree with you on that point. Personally, a single repo is not effective for larger companies and if a repo is the solution it should be git not GitHub. I can’t bring myself in this new age to put our AI ecosystem and IP on any vendor platform, exception to cloud providers infra that I control

[Matthew @MBall_Atlanta]

It’s not just Google and Microsoft, I cannot believe Anthropic and OpenAI have completely missed the boat on this opportunity. Which I’m glad they have anyway because I would not subscribe to the vendor lock in. And I get repos for early tech adopters, absolutely wrong approach for a large company who cares about their business user adoption

[Matthew @MBall_Atlanta]

#3

On BERD in particular, I’m really sad that once I directly engage with an agent that it feels same ol, same ol as a chat or cowork interface. Can you please bring my guy or gal into view, maybe some sounds, doing push ups or something?

This is critical to our business. 😂 okay, I personally desire this.

[Matthew @MBall_Atlanta]

My first call out was I love Berd and only missing the natural switch to multi-player that I love about Buzz.

#2 simple but powerful opportunity, I wish upon building agent instructions I could also select from my list of available skills to give my character a set of competencies. I do it today already but forcing it into one agent instructions box.

[Matthew @MBall_Atlanta]

@morganmartn Music to my ears! Let’s go.

[Matthew @MBall_Atlanta]

@morganmartn @zex_exe @hams I enjoy BERD because I can work in any folders or context like a cowork but I am bummed I can’t talk to coworkers in that same context and collaborate like Buzz, but buzz misses the cowork aspect of picking a place to work in. Can you mate the two real quick like please

[Matthew @MBall_Atlanta]

@hams Having fun with Berd, it’s a lot of what I dreamed of internally…better of course. I just wish I could get Berd multi-player so we could collaborate with teammates simultaneously.

[Matthew @MBall_Atlanta]

@jewelia Having fun with Berd, it’s a lot of what I dreamed of internally…better of course. I just wish I could get Berd multi-player so we could collaborate with teammates simultaneously.

[Matthew @MBall_Atlanta]

@rauchg It’s a fantasy for many companies, whether it’s risk, trust, security, or change adverse…it’s still a fantasy and will be the exception not the standard to think most companies operrate on a mone repo…

[Matthew @MBall_Atlanta]

@AndrewYNg Again excited about they but it doesn’t accept key set ups to LLM gateways - why not add a basic custom integration type?