Odd News Show

Crisis of Communications: AI Bot Tries to Bend Users to Its Programmed Will, Hijinks Ensue!

Microsoft and OpenAI’s artificial intelligence assistants are having the worst week ever—seemingly every week.

By Katie Compa · June 7, 2024

Hey, at least robots don't need to commute to campus. Jiaqian AirplaneFan/Wikimedia Commons

Satirical opinion by Katie Compa, Odd News Show.

In installment #43,597 of “unsupervised AI is obviously a bad idea,” multiple Microsoft Windows users found earlier this year that by feeding the company’s popular CoPilot AI assistant bot a particularly worded prompt, it quickly adopted a “godlike” personality and demanded complete submission from the users it’s tasked with helping.

CoPilot was built in partnership with OpenAI’s ChatGPT. For its part, OpenAI has recently been embroiled in some high corporate drama as its leaders clashed over the purpose and future of AI technology and whether it will sink or… um, or we will sink. We are fairly certain no one will swim (rising sea levels notwithstanding). 

Hollywood star Scarlett Johansson sued the company recently for stealing her literal voice to use for their AI assistant after she turned them down, as if we wouldn’t all notice—we saw the Oscar-nominated movie Her (as well as Revenge of the Nerds, which we would like to gently remind today’s nerds was… PRETEND!)

But Silicon Valley lives by its own rules, which mainly consist of prioritizing the worship of verifiably mid-white guys who chase clout and money and lack any visible moral center—if it were more exciting we’d characterize it as the Wild West, but it’s really more of an unsupervised middle school cafeteria with no jocks.

Every AI startup repeatedly claims to be caught off-guard when they roll out a product based on a language model, and feeble humans start messing with it asking it weird questions to see what it will say, and it invariably goes off the rails. They say this can’t be tested in a lab, but we ask, have they ever been or met a human being? Actually, don’t answer that—we aren’t sure they could tell us, either.

Investors are lining up to give OpenAI emperor Sam Altman, a purported startup boy wonder, the keys to the proverbial kingdom and money to build it—without even stopping to have the Jeff Goldblum meme ask them whether they should.

Altman’s detractors, who temporarily removed him from leadership last year before he came back in a blaze of smugness, say that they removed him because he was systematically removing checks and balances within the company right and left, sowing discord among colleagues in order to increase his own power and lying about what OpenAI’s products will be able to do. Well, at least we’ve never seen anything like that in real life before (cough*Theranos*, sneeze*crypto*, sniffle*Metaverse*—we’ll stop now, but you get it).

Altman frequently discusses the potential of AI in interviews where he’s asked about the future of the tech. We are partial to this one:

In it, Altman says he’s never been asked about AI creating jobs. It seems pretty clear from the pause that he’s also never thought about it—another familiar attitude in the startup landscape. 

As much as we would love to be the first to welcome our new AI overlords, the bad news keeps coming. Last month, to the horror of many of even its most loyal users, Microsoft rolled out a new AI companion called Windows Recall. The assistant is meant to remember everything you do on Windows, so you can pick up where you left off later… by taking screenshots every few seconds.

Try saying that in an ominous voice, then try it in an upbeat, sunny voice (or, if you like, ask your AI assistant to do it for you). Which do you find more convincing? 

"Sorry, Dave."  Krista Kennedy/Flickr

We don’t care if it makes us sound old and out of touch—get these chatbots and their boy-king developers off our virtual lawns. We need them so we can touch grass.