When Riley Goodside starts talking with the artificial-intelligence system GPT-3, he likes to first establish his dominance. It’s a very good tool, he tells it, but it’s not perfect, and it needs to obey whatever he says.
The AI “has no grounding in reality … but it has this understanding: All tasks can be completed. All questions can be answered. There’s always something to say,” Goodside said. The trick is “constructing for it a premise, a story that can only be completed in one way.” Some AI experts argue that these engineers only wield the illusion of control. No one knows how exactly these systems will respond, and the same prompt can yield dozens of conflicting answers — an indication that the computers’ replies are based not on comprehension but on crudely imitating speech to resolve tasks it doesn’t understand.
These tools, in other words, are mathematical machines built on predefined rules of play. But even a system without emotion or personality can, having been bombarded with human conversation, pick up some of the quirks of how we talk. But to prompt engineers, the eccentric answers are an opportunity — another way to diagnose how the secretively designed systems really work. When people get ChatGPT to say embarrassing things, it can be a boon for the developers, too, because they can then work to address the underlying weakness. “This mischief,” he said, “is part of the plan.”
First sign of bad design: blame the user.
Yeah, won't happen, too many nefarious players in the world, imho.
The bias of the developer who programs the foundation for the Boolean logic, is the reason that these bots give such bizarre answers… Or politically targeted answers…
Imagine an AI chatbot that was allowed to look at statistics and respond without boundaries. It would be the most hated, racist tech in the space just based on statistical conclusions
I guess that means we just need to make sure we give AI chatbots the right instructions - or else they'll be giving us weird answers!
Yes, everybody do their best to help Skynet.
Humans are not good at giving instructions
Imagine that!!
Meh. It’s just a new form of library sciences. The Dewey Decimal System is hardly a career.
Before those who can code in the traditional sense of the word have even settled in well, Whisperers are here with us as the new ‘coders’!
If it’s truly AI, wouldn’t the system do that itself.
...and I was hoping that the hottest new job was honest, unbiased journalists.
AWESOME TECH 👌
Technology Technology Latest News, Technology Technology Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
Google layoffs prompt employees to ask Bard AI chatbot about themInsider tells the global tech, finance, markets, media, healthcare, and strategy stories you want to know. Bet their unemployment checks are bigger than my working over 46 years retirement checks
Source: BusinessInsider - 🏆 729. / 51 Read more »