Bing Can Be Bad

The irony of tech people talking for years about how robots are coming for all non-tech jobs and then building AI so that the most likely job to be replaced in the short term are in tech is... hilarious? Am I fired? It was fun while it lasted.

---------

Do I love talking about AI for two weeks in a row? Not especially.
Is it the most interesting thing happening in tech right now? Yes, by far.

So last week we had the big Bing unveiling. Microsoft is cool again. Satya is a hero. Etc.

What also happened last week is that the Bing chat function (the thing built on top of the most advanced version of ChatGPT) was released to a small group of beta users. These are influential people across tech. A real who's who of tech writers, founders, and investors.

There's a waitlist you can join, but I'm pretty sure it's infinite-people long so stories from beta users are probably the closest we'll get for a while. Lucky for us though - some of the beta users are pretty thorough in their coverage.

A couple must reads (if you haven't already):

A Conversation With Bing’s Chatbot Left Me Deeply Unsettled, by Kevin Roose (NYT)

From Bing to Sydney, by Ben Thompson (Stratechery)

The irony of tech people talking for years about how robots are coming for all non-tech jobs and then building AI so that the most likely job to be replaced in the short term are in tech is... hilarious? Am I fired? It was fun while it lasted.

Kind of funny take: The meme of "omg can you believe what ChatGPT just said" is starting to get a bit old. See personal interpretation at top of newsletter. In a lot of ways it feels like one of those elementary school tongue twisters where if you ask someone to say a certain phrase ten times fast it sounds like they're saying "fart" or something else silly. In this case though you're trying to convince an AI program to say something about nuclear war or racism.

Existential take: There's a certain mystery involved with AI given the black-box nature of how the prompt answers actually comes to be that leaves room for a whole spectrum of interpretations. If something is unknowable you can make arguments ranging from pure science to pure magic. In the case of AI, it seems like we're headed in the direction of over-indexing on the amount of magic we suspect is in the black box. It's definitely more clickable to title an article "Holy Shit It's Alive" than "There's A Lot Of Data In There". I wonder when we'll see our first AI led cult...

Investor take: AI is making it way easier to spin up software products. Whether you use it to code or you use it in your product - it can make a startup look like they're moving incredibly quickly. This time compression is going to be misleading to investors who used to be able to derive early startup quality purely off of speed. The "maybe their first idea sucks but holy shit they're pushing new products so fast they're bound to figure it out" mentality seems like it will lose a lot of weight in the years to come. If everyone can go 100mph the question of which direction a team is headed becomes much more important than whether or not they can ship.