Friday Links: Million dollar Bots, AI in Bunkers and AI Product Feedback

Here are this week’s curated links:

• Meta pays celebrities millions to use them as the basis for chatbots. Meta’s new chatbots include Tom Brady, Paris Hilton, and other well-known people. The company apparently paid significant amounts to secure permission to do so. This seems like an unexpected and pleasant windfall for those who were paid. It seems like an interesting new “medium” to explore for stars. You can also build your own chatbot clone/likeness with services like Delphi. This seems like an area where there might be some genuine value in the medium term. No one can be in two places at once, but your knowledge and likeness can potentially be. On the downside, in 5 or 10yrs, the most popular celebrities may not be humans at all but virtual constructions like Any Malu.
• Meta rolls out GenAI creator tools for marketers. The tools allow useful image manipulations such as expanding backgrounds and ad-copy generators (which already existed in Meta’s ad product to some extent). It’s logical that Meta would add tools like this, but it will raise the bar for all advertisers to use these tools. The question will be whether third-party tools like Canva will do a better job than Meta can. Meta’s key advantage might be in using advertising results data to tune recommendations. Something that’s impossible for third-party tools to replicate.
• Launch.AI for product testing: Cambri (Cambri.IO) added an AI layer to their consumer-focused goods market research platform this week. Launch.AI aims to give more nuanced feedback on launch plans and potential impact. It’s hard to tell how much of a step up this is from what Cambri was already doing, but I wonder how much having AI in the loop will reinforce stale market entries rather than making them more interesting and diverse. This also reminds me of CocaCola’s GenAI Coke Flavor. Still some tuning to do.
• A Hit and Run in San Francisco results in a woman being pinned under a Cruise Car. This sounds like a harrowing incident, but not for the obvious reasons. Initial reports implied that a Cruise AI car (with no driver or passenger) ran over a pedestrian and pinned her under the car. Subsequently, it became clear that the woman was first hit by another driver who fled the scene and into the path of the AI car. Her leg was trapped. Police subsequently requested the Cruise car not to move since moving often causes more traumatic injuries. It’s hard to see what the Cruise car could have done differently in this case.
• Tesla is building a “Bunker” in Austin for its new AI system Dojo (May be Paywalled - here is the same info). It is difficult to say that it doesn’t sound dystopian when AI starts to get housed in “bunkerlike” facilities.

Sending wishes to people on both sides of the Gaza / Israel conflict, hoping that sanity prevails and it gets back to peace as soon as possible. Not to mention the Ukraine conflict, which has now completely disappeared from the headlines. As someone wise once said: “An eye for an eye leaves everybody blind.”

Thank you for reading SteampunkAI. This post is public, so feel free to share it.