Yesterday, I was perusing the photos on my iPhone when I stumbled across a folder called Receipts. That surprised me, as I’ve used Expensify to capture bills while traveling since it came out.
So what was this mysterious receipt?
The photo above contains words and numbers, but it isn’t a receipt. Rather, it’s a playlist from an excellent Marillion concert I attended in Montreal a few years ago. (I met the lucky man who landed this keepsake after the show. As he proudly told me, he’s got a bunch of them.)
AI doesn’t know the difference between right and wrong.
Techies know what Apple is doing here. Since at least 2021, the company’s on-device machine learning has attempted to recognize people, places, and things and then categorize them.1 Sometimes it’s right. As we see above, sometimes it isn’t.
What You Need to Know
While genAI is enormously helpful in certain ways, it suffers from significant limitations:
- It doesn’t know the difference between right and wrong. It cannot tell the difference between lies and truths.
- It doesn’t know if it’s violating fair use—and you won’t either.
- It can’t tell you if its output is original or interesting. (It’s often synthetic.)
- In the event of a grievous error, the tools and their makers aren’t accountable. OpenAI chief Sam Altman won’t take responsibility if ChatGPT fumbles the ball.
Food for thought the next time you decide to let AI do your writing for you.
Footnotes
- Visual Look Up expands this feature.
0 Comments