
https://nitter.gq/t.co/Wuaj6X74Es https://nitter.gq/t.co/wv9Zsd0jLJ AI SEO, https://nitter.gq/t.co/PQQDvRVfgQ, @ColonyNFT, @TribesNFT, Image, Jon Peddie, MAXIS, Studio Cutie
Portland, OR
Joined June 2009
- Tweets41.9K
- Following5.6K
- Followers3.8K
- Likes261K
Now I'm building crazy shit. LOL
I'm letting agents declare their expectations for a search before they send the query and then have the results scored by an outside judge as to how closely it matches.
31
The only bad part about being in the middle of a zeitgeist is there are A LOT of papers I have to read now. And LLM summaries are not going to cut if for these.
25
If one version of AI alignment is based on the "shutdown problem" how is that different from a human alignment problem where you point a gun at their head? That seems less than ideal.
1
1
23
I really need to figure out a way to make money so I can spend 100% of my time on research.
I'm just weeks away from solving agent authority issues and launching products.
5
7
37
AGI is *so* 2025. Memory is 2026. retweeted
The people yelling we need to listen to them because they hate Hakeem Jeffries and Chuck Schumer have been ignoring us when we've been telling them we hate Bernie Sanders for ten fucking years.
This thing with replacing Schumer reeks of pushing Biden out and that got us four years of Trump.
No thanks.
3
62
Before we can solve the AI Alignment problem we still need to solve the HUMAN alignment one.
A good starting point is maybe not destroying humanity.
But beyond that we have no agreed upon ideal society or living conditions humans must adhere to.
So how can AI know what's best?
2
39
AGI is *so* 2025. Memory is 2026. retweeted
startup org chart 2026:
CEO
↓
Claude
↓
more Claude
↓
one intern who knows where the API key lives
Best framing of the trust problem I've read this year!
Agents require responsiblity for their actions to truly be trusted. It may require a different kind of AI tool that lasts.
Nobody trusts your agent.
If 2026 was the year of agents, 2027 will be the year of principals.
An agent acts on someone else’s behalf. A principal bears the consequences.
I wrote an essay on this topic, including what to build and where to invest.
Let me know if you think @boardyai is the first AI principal...
2
72
And nobody on the Left thinks Bernie Sanders is a hypocrite? A "new bold direction"? REALLY? Be honest. How long has *he* been in Congress?
30
I need to design a game that's a combination of Slay the Spire for card mechanics, Risk for land conquest, and a Rogue-like for level advancement and repeat playability.
1
1
23
AGI is *so* 2025. Memory is 2026. retweeted
We are less than a week past the biggest Democratic electoral victory in half a century.
And the people calling for replacing Schumer and Jeffries are dividing us.
That says everything about them you need to know about those people.
AGI is *so* 2025. Memory is 2026. retweeted
🔥🔥🔥Oops I Did It Again:
How Political Science Predicted the 2026 Midterms
On September 8, I put Bryan Steil’s Wisconsin district on my list of 20 Republican seats Democrats would flip. Inside Elections had it Solid Republican. Nine days later, it moved the race to Likely Republican. It took until October 29 to get to Lean Republican. That’s where Cook and Sabato finished, too. Then the voters of southeastern Wisconsin elected Mitchell Berman.
I went 20 for 20.







