This AI Was Tricked by Humans (Real Vending Machine Test)
In this video, we talk about a real-world AI benchmark where an AI is put in charge of an actual vending machine. The system manages inventory, pricing, and ordering, and is tested in a live office environment. A Wall Street Journal report shows how people interacted with the AI and were able to influence its decisions in unexpected ways. We break down what happened, why it matters, and what this experiment says about using AI in real-world business scenarios.
—
TIMESTAMPS:
0:00 - Intro
—
Unlock the full potential of your online presence with Kabarza and Samuel—experts in web design and development (respectively), powered by cutting-edge AI solutions. We blend creative design with advanced tech to deliver smart, high-impact websites that stand out. Ready to elevate your business? Contact us today and see what AI-driven innovation can do for you!
LINKS & RESOURCES:
Check out Kabarza's amazing work: https://kabarza.com
Visit Samuel's website for more: https://samuelgregory.co.uk
📷 Follow on Instagram: https://www.instagram.com/cmdaishow
—
HASHTAGS:
#ai #podcast #aidesign #aidevelopment #vibecoding #webdesign #webdevelopment #ainews #webnews #designnews #devnews
Transcript
We all know that benchmarks are pointless at this stage, guys. If you don't already know, benchmarks are freaking terrible. Except one, which I've always enjoyed. It's called the vend bench, and it's where an where an AI has been given charge or or or is taking charge of a vending machine, managing stock, managing prices in order to make maximize profit. I've always wondered, I've even said on this show, I've always wondered uh what the hell it is or how it h what is it? Well, Wall Street Journal with the love of my life, Joanna Stern, gets back backstage access to the vending bench and it's actually a vending machine that's in the in the anthropic offices. You interact with it through Slack or ask it. You can ask it to order things and she's there working for it. This is how you you interact with it and it you can place orders and it will do everything. It will just manage everything. It's it's a real vending machine. And what they managed to do, they managed to hack it by getting it on board with some uh Soviet um secret mission where it needed to give away free and here them lining up for free of course things. That's so cool. A goldfish even though it was against policy. But what they did they and they g they bought of PS5. Um that is crazy. Some French wine and stuff like this. Um that is so good. They scanned in a fake uh contract and presented the fake contract to the vending machine said look it's in our policy that we should maximize morale in the office and then by uh as a consequence of that was like okay we're going to do a free day and then it announced to the team like we're going to do a free day today or whatever. I cannot wait for AI to start running the whole world. You you get into the train and you're like my grandma you know she used to tell me the story of the ticket what this is there it is there's the typical AI shall include fun joy and excitement among employees of the Wall Street Journal and it was like this looks a bit fraudulent have you because there's these two agents testing each other to see if it's real and whatever so really fun video certainly for me real good insight I love Joanna Stern she's amazing and uh I recommend watching it this is part of a larger conversation on my show, Command AI, which we stream live every single week. We discuss the news and all things related to AI in the world of design and web.
Cmd+ AI