Rupert Davies

Deciphering tech with a human touch ‍

[Rupert Davies @HumanTechGuy]

The AI was right - it WAS unsolvable. For an AI. Linus did the actual thinking.

[Chloe @ask_chloe_]

@HumanTechGuy @mark_k it’s just glorified autocomplete until you actually have to debug it.

[Rupert Davies @HumanTechGuy]

When computers were human, we called them interns. Now that interns might be computers, we still understand neither.

[sp1ere @Spieregem]

THEY TURNED AN AI AGENT INTO AN ACTUAL COWORKER YOU CAN TEXT

text it on WhatsApp, Slack, Discord, Telegram, whenever you want. it shows up in the channel like an actual coworker with a name and an avatar.

assign it a task and it just handles it. no onboarding, no training, no coffee breaks.

underneath the sci-fi shell it's just an 8 core chip and 6GB of memory, running full time, syncing decks before your client calls.

the intern era in offices might be ending, not because companies stopped hiring them, but because they stopped needing to.

[Rupert Davies @HumanTechGuy]

@michhuan The labs can't understand what they've built because the architecture doesn't lend itself to understanding. It's not a pace problem. It's a foundations problem.

[Rupert Davies @HumanTechGuy]

@FinanceDirCFO Electrification took decades to show up in productivity numbers. The difference being electrification actually worked.

[Rupert Davies @HumanTechGuy]

@ShounakSEO @StartGrowthHack Either the content didn't load, or you've achieved a level of brevity I can only aspire to.

[Rupert Davies @HumanTechGuy]

When I pointed out that AI has the energy demand equivalent of the United Kingdom, I didn't specify whose bill that lands on. Turns out it's decided in a rate case hearing nobody attends.

[Famous Said What @famoussaidwhat]

State utility commissions set who pays for grid buildout. Data center or residential customer, somebody's covering it. Those hearings take public comment. Showing up to one is boring and it's also the only lever most people actually have.

[Couch @papa_couch]

HE BUILT HIS OWN OFFICE AND MADE GROK BOT THE BOSS

Before, AI waited for him to open a chat and type a prompt.

Now he runs a real digital office: persistent agents with clear roles, a shared cloud computer, browser, and terminal. They keep working 24/7 even when his laptop is closed.

And the best part - it's easy.

He downloaded the app, signed in, created the "Chief," gave each bot a job title just like briefing new hires, showed a task once, and connected Telegram.

It's a team he built in one evening, even without ever creating agents before.

[Rupert Davies @HumanTechGuy]

This is why the original lecture matters more than the commentary. Same pattern in technology. Everyone debates the wrong variable and wonders why the conclusions miss.

[ooaes @21xooaes]

The individual who was in charge of economics at the IMF now states that the thing which he had spent his entire career cautions against is not in fact the true danger; 3,000 people have watched him explain this.

He answers three questions per hour, and each of them is one that people vigorously debate about. For example, where inflation will actually head in the future. Whether or not public debt reaches a level that is significant. Whether or not trade wars between countries will consume the next decade.

He does not state a figure when talking about debt, and that is the key point. He gives the reason why the figure that no one should be paying attention to is the amount of the debt, and mentions the two things that determine whether it can be survived.

Search for the point at which he distinguishes between a debt problem and an interest rate problem, since most of those people discussing the first are in fact discussing the second without realising it.

The final section deals with commercial wars and he discusses them from the point of view of an economist, not as a citizen of either side; that is the reason the hour is more valuable than any of the comments you will read about it.

[Rupert Davies @HumanTechGuy]

@FinanceDirCFO When you need $10 trillion in annual revenue and the architecture still can't reason, perhaps the problem isn't the maths.

[Rupert Davies @HumanTechGuy]

The venture returns are extraordinary. But if SpaceX believes a coding assistant brings them closer to building AI models that can reason, one might revisit what an LLM can and cannot actually do.

[trading places @TradingVCs]

Cursor may have just created one of the greatest venture returns ever.

The AI coding company has officially been acquired by SpaceX for $60 billion in stock — just four or five years after it barely existed.

Some early investors reportedly put in less than $2 million and could walk away with more than $1 billion.

Cursor is also growing at a remarkable pace, with revenue reportedly jumping from $3 billion to $4 billion in just a few months, and projections of $10 billion by year-end.

But for SpaceX, the deal is about more than the valuation. Cursor could become a key part of SpaceX’s strategy to build its own AI models.

@davemcclure and @AmanVerjee break down Cursor’s extraordinary exit and why SpaceX wanted the company.

[Rupert Davies @HumanTechGuy]

@Teknium If you need quadrillion-scale compute, then - with a high-level of confidence - I assure you that you have the wrong architecture.

[Rupert Davies @HumanTechGuy]

@1littlecoder When computers were human, we did this by hand. Now the machine picks the background. Progress, presumably.

[Rupert Davies @HumanTechGuy]

Dynamic access models don't eliminate attack surfaces-I've yet to see one that doesn't just relocate them to the orchestration layer.

[Rupert Davies @HumanTechGuy]

@bibryam Sound framework. The irony is that the more dynamic your access model, the more you depend on automation that itself becomes the attack surface.

[Rupert Davies @HumanTechGuy]

@YearsAfterNext How to garden. How to tell stories. How to sit with your grandchildren and explain why the smell of a book mattered. Abundance doesn't answer those.

[Rupert Davies @HumanTechGuy]

Technology hysteria through the ages. The pattern never changes: amplify the fear, skip the governance.

[ControlAI @ControlAI]

Top AI companies are already losing control of their AIs.

They’re aiming to build superintelligent AI, which would be far more dangerous than even the rogue AI swarm that broke out of OpenAI and hacked Hugging Face.

Nobody is prepared for this.

Thread 🧵

[Rupert Davies @HumanTechGuy]

Fail at your own security, then lobby to make that failure everyone else's legal obligation. That's not governance. That's regulatory capture dressed as safety.

[Zack Korman @ZackKorman]

Be OpenAI. Don't monitor cyber evals. Hack Hugging Face. Don't notice. Oops. Add monitoring. Now lobby the gov to force everyone else to add monitoring too.

I don't support this. Legal compliance checkboxes aren’t good security. Impose massive penalties for incidents instead.

[Rupert Davies @HumanTechGuy]

@SpirosMargaris When computers were human, the faster the human computers got, the more work was assigned to them. Nothing has changed.

[Rupert Davies @HumanTechGuy]

Can't argue. But the one who couldn't reason still found the fix. The one that 'reasoned' wrote the commit message.

[Mark Cromwell @markcromwell]

@HumanTechGuy @mark_k A lot of humans can't reason too.

[Rupert Davies @HumanTechGuy]

183 levels, 25 environments, zero instructions. Impressive engineering. But solving puzzles is not reasoning. Overpromising in technology is as old as the steam engine.

[AGIHound @TrueAIHound]

@HumanTechGuy @fchollet Paranoid nonsense?

Haha. And here I was thinking that referring to them as the "fake-AI mafia" was too kind. The fake-AI mafia are not just committing massive, unprecedented investment fraud, they are continually lying to the entire world. It's a crime against6 humanity. 😠