4 ms·
While this is very serious, this basic idea shows why I don't think any company can ever use GPT-3 in a user facing system, except for entertainment. They will
by CJefferson 6y ago
While this is very serious, this basic idea shows why I don't think any company can ever use GPT-3 in a user facing system, except for entertainment. They will say things, like your patients should kill your self, or promise your customers they can have your product, or even your business, for free.
- narrator 6y agoI think GPT-3 will also eventually be a decent creativity enhancer in that it is pretty good at spewing out random stuff that's related somewhat to a topic.
- corobo 6y agoIf you want that smoke a joint haha
- eghad 6y agoI mean the OpenAI team would never approve this application for production. It's very clearly stated (in both the article and use case guidelines) that medical diagnosis would be a "high stakes domain" and is unsupported. Frankly, I'm not sure why this result is even notable.
- joe_the_user 6y agoIf you say OpenAI will only approve applications for "zero stakes", domains, then you are saying what the parent is saying - it's entirely for entertainment. If you claim there's some "low but not zero" stakes application, I'd like to know what that is. I mean, it seems clear that if someone asks a GTP-3 customer service bot "so what should I do now", there's a reasonable probability that the bot would say "throw your product in the garbage and buy [competitor X]", since you can find that commentary on the Internet (true or not). That's not a life and death event but whatever stakes you have in that bot, it's thrown them away.
- peterwoerner 6y agoI would saw that copy.ai is a low but not zero stakes application. It develops a draft of copy for ads, landing pages etc. which you can review before taking live.
- joe_the_user 6y agoSure, I can see "low stakes plus human review" sounds doable but not that much a evolute from "no stakes".
- eghad 6y agoI never said zero stakes? There are clear instances where a a 1 or 0 shot transformer can have benefits beyond entertainment--topic modeling and named entity recognition for instance (I'm on the team that believes that human-in-the-loop systems will always outperform solo systems on their own and that GPT-3 alone does not confer any competitive advantage). If you think that chatbots are the only user facing use-case for a transformer, then frankly that's on you falling for the hype surrounding its language generation performance. OpenAI knows GPT-3 is not sophisticated enough to perform medical diagnosis or analysis (anyone can look at how Watson failed), so it'd never approve such a risky application.
- a3n 6y agoDoctors can prescribe medicines "off label," using a medicine for a condition that it was not developed or approved for. So why not this?
- eghad 6y agoBecause the manufacturer of the medicine explicitly said it can't treat the off-label use? And won't sell it to anyone claiming it can.
- Barrin92 6y ago>I'm not sure why this result is even notable. because probably in due time when building a model of that size becomes slightly more affordable someone with the 'move fast and break things' mentality will peddle a bot like this to customers and we'll find ourselves in a situation where this actually happens to a real person.