Alberto Romero

1 week ago | medium.com | Alberto Romero

Would you have fallen for the emotional support kangaroo? As a wise man once told me, the trouble with AI won’t come from the impossible, but from the almost-true, always lingering on the fringes of reality. The first high-quality AI video has gone viral, in the proper sense of the word: It has reached and fooled millions of people. It features a cute “emotional support” kangaroo being denied access to a plane. As it ends, the video zooms in on the kangaroo and shows him holding the boarding pass.

1 week ago | thealgorithmicbridge.com | Alberto Romero

As a wise man once told me, the trouble with AI won’t come from the impossible, but from the almost-true, always lingering on the fringes of reality. The first high-quality AI video has gone viral, in the proper sense of the word: It has reached and fooled millions of people. It features a cute “emotional support” kangaroo being denied access to a plane. As it ends, the video zooms in on the kangaroo and shows him holding the boarding pass.

1 week ago | thealgorithmicbridge.com | Alberto Romero

When a standard software program doesn’t obey your commands, you restart your computer and go on with your day. When an AI software program does it, you might end up dead. Or worse, blackmailed. That’s what Anthropic researchers found about the recently released Claude 4 Opus and what AI safety organization Palisade Research found about OpenAI o3 models: they won’t comply when threatened to be shut down. In the case of o3, not even when explicitly instructed to allow itself to be shut down.

1 week ago | medium.com | Alberto Romero

OpenAI and Anthropic make the puppets, but the puppeteer could be anyoneWhen a standard software program doesn’t obey your commands, you restart your computer and go on with your day. When an AI software program does it, you might end up dead. Or worse, blackmailed. That’s what Anthropic researchers found about the recently released Claude 4 Opus and what AI safety organization Palisade Research found about OpenAI o3 models: they won’t comply when threatened to be shut down.

1 week ago | thealgorithmicbridge.com | Alberto Romero

AI has already stolen enough from us—don’t let it take the em dashes too. I keep running into writers who go into fight or flight mode when faced with the bleak prospect of an AI writing avalanche. They either never use an em dash again, the semicolon, or the Oxford comma; or double down on them—and let’s see if someone dares accuse me of being AI!I think both choices are a mistake. The natural kind, sure, but a mistake nonetheless.