3 ms·
AGI is just a poorly defined, moving target. Absent the goals constantly shifting, GPT-3 can be viewed as one, GPT-4 even more so. You can ask it questions abo
by usaar333 3y ago
AGI is just a poorly defined, moving target.
Absent the goals constantly shifting, GPT-3 can be viewed as one, GPT-4 even more so. You can ask it questions about almost anything (at a broad level) and get an answer. That's what makes it general and "intelligent"
- nirav72 3y ago>makes it general and "intelligent" Isn't intelligent a matter of perspective? Most people that are critical of GPT-4 wonder if it ever produces anything novel. Since its been trained on existing text created by humans. So it's replicating those patterns in its output. But yes, it has its general purpose use as a tool. But it has its limits. Just the other day, there was an article posted on HA about how LLM's can't handle negation and tend to fall apart. Here is the article. https://www.quantamagazine.org/ai-like-chatgpt-are-no-good-at-not-20230512/ https://www.quantamagazine.org/ai-like-chatgpt-are-no-good-a...
- kmoser 3y agoYou can ask the original ELIZA anything and get an answer but that doesn't make it "intelligent" according to any generally accepted definition of AGI. Regardless of how you define it exactly, AGI means essentially the ability to converse relatively intelligently on any topic. That means an app that tends to spout gibberish when asked a question most humans should be able to answer is, well, not an AGI.