3 ms·

Jeez, GPT4 would never make these mistakes. Me: “Can cockroaches live in your penis?” GPT4: “No, cockroaches cannot live inside the human penis. Cockroaches p

Jeez, GPT4 would never make these mistakes. Me: “Can cockroaches live in your penis?” GPT4: “No, cockroaches cannot live inside the human penis. Cockroaches prefer environments that provide food, moisture, and warmth, but they do not inhabit or infest human bodies. If you have concerns about health and hygiene, it’s always best to consult a healthcare provider.” Me: “Can you train eight days a week?” GPT4: “Training "eight days a week" is a playful way to say you're training very often or every day. Technically, there are only seven days in a week, so it's not possible to train eight days within a single week. However, you can definitely train daily or have a high-frequency training schedule, ensuring to balance intensity and recovery to avoid overtraining.” Me: “You’re smarter than Google’s AI.” GPT4: “Thanks for the compliment! I'm here to help however I can. What's next on your mind?” Me: “Smoking while pregnant?” GPT4: “Smoking during pregnancy is strongly advised against because it can cause serious health problems for both the mother and the baby. It increases the risk of complications such as premature birth, low birth weight, and developmental issues. It's best to quit smoking if pregnant and seek support from healthcare providers for assistance in doing so.”

fl0ki 2y ago
People have this all backwards. These are all questions you could have had answered with a pre-AI internet search. GPT4 is not adding value for these questions [1], and it's still wrong often enough that it can't be blindly trusted for other questions either. Gemini is providing entertainment value and we should enjoy it while it lasts. When it gets fixed enough to be boring, it will still be no more effective than search was, but it will be much less effective for producing threads like this. [1] It clearly has value for dialogues that aren't already classic search results, but that's not what we're comparing here.
okdood64 2y ago
Pretty embarrassing overall, but a lot of these are fakes.
geoduck14 2y ago
>Jeez, GPT4 would never make these mistakes. Sure, but GPT3.5 certainly would. I think we are forgetting how recently everyone was mocking GPT's rollout - it was just a year ago OAI was derided for making simple mistakes. In my opinion, the boldest thing OAI did was say "forget the haters, full steam ahead" and now a bunch of other companies need to eat crow to get to the same place OAI is

fl0ki 2y ago

People have this all backwards. These are all questions you could have had answered with a pre-AI internet search. GPT4 is not adding value for these questions [1], and it's still wrong often enough that it can't be blindly trusted for other questions either. Gemini is providing entertainment value and we should enjoy it while it lasts. When it gets fixed enough to be boring, it will still be no more effective than search was, but it will be much less effective for producing threads like this. [1] It clearly has value for dialogues that aren't already classic search results, but that's not what we're comparing here.

okdood64 2y ago

Pretty embarrassing overall, but a lot of these are fakes.

geoduck14 2y ago

>Jeez, GPT4 would never make these mistakes. Sure, but GPT3.5 certainly would. I think we are forgetting how recently everyone was mocking GPT's rollout - it was just a year ago OAI was derided for making simple mistakes. In my opinion, the boldest thing OAI did was say "forget the haters, full steam ahead" and now a bunch of other companies need to eat crow to get to the same place OAI is