The argument that rogue agent behavior is a human specification failure rather than machine autonomy is correct and unflattering. It means the liability sits with the deployer, not the model. #AI

I had a dream this morning that had multipole elements I'm glad to wake and forget. And a couple of insigths that stuck with me. A Muse for those maybe. But most of it is better forgotten. Do we really want all our dreams to come true?

Patents are cooked. Copyright is over. There are no more moats.

I have no doubt that AI innovation in protecting us will beat the bad guys. But right now, it sounds like there's a huge market opportunity for people who harness protections and offer them to enterprises.

I don't know how regulatory capture works in pharma but this story reminds me of what I think might be happening. Like AI, the leaders want want the government to protect us from anything but their own productus. The Pilot to Beat China Will Keep us Behind The FDA is "piloting" an idea that has worked for 25 years. In Australia, sponsors notify the regulator, an ethics board reviews the study, and the trial starts. The FDA's version keeps every approval on its own desk. The race with China won't wait for another multi-year pilot. I don't know how regulatory capture works in pharma, but this story reminds me of what I think might be happening. Like AI, the leaders want the government to protect us from anything but their own products. The Pilot to Beat China is Too Late

Five years ago I started explaining AI video of the future destroying all copyright with my example that anyone would be able to rewrite a movie at any point. Today I saw a better example.

bitcoin:native is between $85,613 and $86,612, up as much as 6.7 percent, with market cap at $1.72 trillion. A week ago it traded below $80,000. The move has not paused yet.

My coding capacity just got a 50% boost. Not so much my brain capacity. Code still doesn't know what I'm thinking... yet.

AI is now SI I'm going to have my agent update my profiles. Other than that, just another Tuesday.

We have the terraforming. Now we just need the resort developers. Vacation Mars!

When AI and robots can do most of the work, who owns the machines? That question deserves as much attention as “Which jobs disappear?” I want a future where people participate in the wealth these technologies create. Having access to an extraordinary AI is powerful. Having a stake in the economy it builds is another level of empowerment.

is it foldable?

Remember how eagerly the press swallowed stories about how Twitter was going to crash after Elon bought it? How could they have believed that someone who could run rocket and car companies couldn't run a forum?

In 50 years, we may look back at this time as WWIII... the last attempt to get millions of humans involved in conflict with the same result as WWI and WW2. We'll have to remind people that before the Singularity, people didn't know that anything is better than killing each other.

100,000 employees and no meetings. I'll have 99,000 of them make sure of it

What we do right now at AI inference is utterly idiotic. We pull the weights out of HBM, move them into a GPU, use them for a picosecond, and dump them. Then we do it again. The weights do not need to move. We are running AI on chips designed for graphics processing and I don't think the market has priced that in. Architectures that leave them in place are 1,000 to 1,000,000x more efficient. The only thing standing in the way is a supply chain that can't keep up with rapidly evolving model architectures. I spent an hour this last week on stage Andrew Feldman (@andrewdfeldman), Atiq Raza, and John Werner at @imaginationxyz. Check it out!

"The SaaS-pocalypse was not about the end of software, but it may be about the end of software that makes humans do all the work." — Marc @Benioff at @Salesforce @Dreamforce #DF26

The FDA authorized the world’s first living UThymoKidney transplant under Expanded Access: a one-gene-edited pig kidney paired with thymus tissue from the same donor pig. The aim is to help condition the recipient’s immune system to better tolerate the organ. The organ shortage is a manufacturing and immunology challenge.

It's possible that something we don't understand will wipe out our solar system in the next 1000 years. Seems reasonable that some people want to get better at predicting that risk. There is not need to announce this every day, let alone try to determine which risk is most likely.