[crote]

The Anthropic ads are heavily into the uncanny valley. The services they are demonstrating looks horrible - even before the ad.

A soulless psychiatrist who'll give you generic cookie-cutter advice about deeply personal issues? Why would you want that!?

Same with the personal trainer, the startup coach, and the professor. Any of them would be incredibly creepy in real life, with their fake smiling, uncanny repeated stock phrases, and fake positivity.

They are trying to spin it like the integrated ads are the problem, but the services are too far detached from genuine human behaviour for that to matter. "Our creepy ripoff psychiatrist doesn't have ads" isn't exactly a great message, is it?

[krackers]

Aside from the ads/no-ads, they're also trying to lampoon chatgpt (especially the "sycophantic" 4o style), since Claude is supposed to be a more "human" LLM (or at least Anthropic likes to think so given their focus on constitutional rlaif, "soul doc" and whatnot).

But it's not a good ad when the only people who will get the reference are those plugged into "ai twitter". But association by implication doesn't work, the only thing most people will end up associating is the creepy guy with "Claude"

[tzs]

> But it's not a good ad when the only people who will get the reference are those plugged into "ai twitter". But association by implication doesn't work, the only thing most people will end up associating is the creepy guy with "Claude"

It might also work for people who watch "South Park". I've never used any LLM speech interface, and until recently had only ever asked ChatGPT short one off questions and so had never seen the sycophantic tendencies, but immediately recognized the creepy people in the ads were supposed to represent ChatGPT from its portrayal in the South Park episode "Sickofancy" from August of last year.