Starbucks controls their supply chain. If they wanted to, they could have their manufacturers put RFID tags on sleeves of cups, bottles, etc and automate inventory with perfect accuracy. Instead, they spent way more money on an AI solution, then had to abandon it because it doesn’t work.
I constantly (although, actually less so recently) get asked if I can do something and I say “yes” before they finish the statement with “with AI”.
The end goal is making a LinkedIn post… not about whatever they asked.
Just do what everyone else is doing and lie
I’m seeing some applications for it as an easy-to-implement context-aware OCR scanner for things that have consistent content in wildly inconsistent formats. Think lists of open invoices from suppliers in a thousand different for-printing layouts, plus some just being screenahots from their systems. The AI system can process those and output a consistent list, which can then be matched against a database the old-fashioned way.
That’s the only practical use I’ve found.
I had the opportunity to listen to professionals from Microsoft and another huge LLM integrator presenting their solutions. Both had no problem admitting you can’t use LLMs for cases that require accuracy or predictability, they were quite honest about it. How one can even attempt to handle inventory management with this is beyond me, this project was designed and implemented by fools.
How long before they admit that these things are garbage, and can’t be fixed?
However long it takes for the funding to run out and the bubble to burst.
👍
