Hey, AI Insiders! It’s summertime, and the AI is easy! For once, things seem sleepy in the AI verse this week. So, we’re changing up and offering just a few news mcnuggets, starting with an update about McDonald’s. And I’ve given the bots time off this week. The only byline today is mine. Enjoy!
Brief Bytes
By Michelle Johnson, Editor, AI Insider
Hold the AI. McDonald’s has put the skids on an AI drive-thru ordering system they were testing in about 100 locations. The company is rethinking the setup and says it will make a decision about the next steps by the end of the year.
Targeting AI: Meanwhile, Target is putting AI into the hands of its employees to empower them to assist customers. According to this New York Times piece (no paywall), At Target, Store Workers Become A.I. Conduits. “Target is testing the device in 400 stores and plans to make the app available to most workers across its nearly 2,000 locations by August,” the Times piece said.
Updated Claude Bot Drops: Anthropic, maker of the Claude bot, has dropped a new version, Claude 3.5 Sonnet, and as usual, some of the headlines focusing on the horse race are just plain annoying. Every time I see a headline like this, I cringe. It’s meaningless because you just know that next week, the week after, a month after, the headline will read: “ChatGPT beats the stuffing out of Claude.” Seriously, this isn’t helpful info for the average user.
Here’s what we actually need to know: If you’re a programmer, you’ll enjoy Claude 3.5. It’s apparently better at coding. If you use Claude for other stuff, you’ll find that 3.5 runs faster, and there are reports that it now has more of a sense of humor.
The “vision” feature, which looks at stuff and answers questions about it, is apparently really well done.
Just like with ChatGPT, there’s a free version, but you can only enter a limited number of prompts before they try to get you to pay.
Superintelligence Superheroes? Ilya Sutskever, a co-founder of OpenAI (ChatGPT), who recently walked away from the company amid charges that they’re not focused enough on safety, has formed a new company called Safe Superintelligence Inc. There aren’t many details about it yet, including who’s backing it financially. Hmm. Wonder who they were referring to when they said in their launch announcement: “Our singular focus means no distraction by management overhead or product cycles, and our business model means safety, security, and progress are all insulated from short-term commercial pressures.” Cough: “OpenAI!” Cough: “Microsoft!”
AI in the Pipelines: What gets lost in the hype about AI sometimes is the inroads that this tech is making in practical applications. For instance, we know it’s making inroads in medicine, but that’s not the focus of much coverage. Well, here’s something that isn’t getting much airtime. TechCrunch says: SewerAI uses AI to spot defects in sewer pipes. Jokes aside, this story notes that “Climate change is raising the risk, rate, and cost of sewage failures. Floods are becoming more common, leading to backups that frequently overwhelm wastewater treatment systems.”
And, they add, the country’s failing infrastructure doesn’t make it better.
So, a couple of guys in the industry founded SewerAI to help. SewerAI was trained on video “footage of inspections of 135 million feet of pipes from municipalities and independent contractors.” That means it can forecast future problems, which can help managers focus on what to fix next.
This clearly isn’t as sexy as generating fancy art or chatting with you about the nature of the universe. But admit it. You’ll appreciate this kind of AI when your pipes don’t back up into a smelly mess.
What. The. Actual. Hell.
File this under “Sadly Clueless.”
Like many similar fan communities on the Discord group chat platform, Midjourney, my favorite AI image generator, has a feature called the “Daily Theme.” Every 24 hours, they invite users to write a prompt that will generate an image related to a theme and share it with the masses.
It’s a fun feature that showcases the creativity of humans worldwide in a feed that all members can scroll. The themes are mostly generic, covering anything from fractals to fog. One user answered the fog challenge with this prompt: “Golden fog flows through a dark valley as if it were a milky river.” (Creative, right?)
On June 19, aka Juneteenth, guess what the Midjourney Daily Theme was? Two words: “strange fruit.” Yes, really. If you don’t know why that’s a problem, Google “Lyrics Strange Fruit.”
Yeah, they did this on Juneteenth.
The theme is usually up for 24 hours, but before the rollover, it was changed to simply “fruit.”
Before that happened, I may or may not have generated an image with this prompt: Billie Holiday sings Strange Fruit before a clueless white audience on Juneteenth. (Sadly, it ignored my request. The audience was melanated. Anyway….)
Someone else’s prompt summoned an image of Nina Simone singing the song. User @Demetria asked: Did the connotation of strange fruit finally dawn on y’all?“ after the prompt was changed.
This incident demonstrates why education is a terrible thing to waste. I’m looking at you, Florida. And, you tell ’em @Demetria.
Aht* Gallery
The dictionary definition of a “humanoid robot” is a robot that resembles a human, sometimes referred to as an android. Here’s how Midjourney interprets the prompt: “A humanoid robot with African American features, typing on a computer.” How did it do? (*Because, Boston.)
