Testing Chat GPT by asking science questions.
Testing Chat GPT by asking science questions.
📫𝐎𝐮𝐫 𝐅𝐁 𝐏𝐚𝐠𝐞:
/ scienceworld-106933907791981
ChatGPT made by OpenAI (https://openai.com)
📚𝐃𝐚𝐯𝐢𝐝'𝐬 𝐁𝐨𝐨𝐤𝐬
📕 𝗪𝗲𝗶𝗿𝗱 𝗠𝗮𝘁𝗵𝘀: 𝗔𝘁 𝘁𝗵𝗲 𝗘𝗱𝗴𝗲 𝗼𝗳 𝗜𝗻𝗳𝗶𝗻𝗶𝘁𝘆 𝗮𝗻𝗱 𝗕𝗲𝘆𝗼𝗻𝗱
(https://www.amazon.com/Weird-Maths-Ag...)
📙 𝗪𝗲𝗶𝗿𝗱𝗲𝗿 𝗠𝗮𝘁𝗵𝘀: 𝗔𝘁 𝘁𝗵𝗲 𝗘𝗱𝗴𝗲 𝗼𝗳 𝘁𝗵𝗲 𝗣𝗼𝘀𝘀𝗶𝗯𝗹𝗲
(https://www.amazon.com/Weirder-Maths-...)
📗 𝗪𝗲𝗶𝗿𝗱𝗲𝘀𝘁 𝗠𝗮𝘁𝗵𝘀: 𝗔𝘁 𝘁𝗵𝗲 𝗙𝗿𝗼𝗻𝘁𝗶𝗲𝗿𝘀 𝗼𝗳 𝗥𝗲𝗮𝘀𝗼𝗻
(https://www.amazon.com/Weirdest-Maths...)
** The kindle versions are available
*** For more details : http://weirdmaths.com/
📄𝗧𝗿𝗮𝗻𝘀𝗰𝗿𝗶𝗽𝘁𝗶𝗼𝗻:
If you haven’t heard yet, ChatGPT is an AI chatbot that can write codes and answer any type of question. It doesn’t have Internet connection, so all answers it gives, are produced by AI itself. Some people consider that it might eventually take over from Google.
We’ve made a lot of science videos for this channel. So let’s ask some scientific questions and see how ChatGPT responds.
Sometimes we’ll deliberately make grammatical mistakes and write typos just as people do all the time. We want to see, if the AI will be able to understand us.
Let’s start with a straightforward question: What’s the farthest galaxy?. A new candidate has been found but ChatGPT has got it correct.
Next let’s ask, as in one of our latest videos, why are there no green stars? Most of the information is correct, but then it says “the colors stars emit do not include green”. Actually this is wrong. Some stars emit all colors, including green, but we see them as red, orange, yellow, white or blue. But the answer is still promising.
Now let’s ask a difficult one: who invented the first seismoscope?. Wow, we thought it was going to say “I don't know” but it does know. It’s even written about the working mechanism. Impressive.
Let’s ask about the Fermi problem. We think it’s really likely to fail with this one. But, surprise, surprise, it knows what Fermi’s method is. But its estimate is too high. Let’s ask again. It’s getting closer. We think it’s done a good job. The actual answer is 83. Will it continue to learn? Let’s ask again . No it hasn't made any more progress. It seems as if it may have to come across this question quite a few times before it learns be more accurate. Let’s move on.
Let’s ask another difficult question. During the Second World War, the Japanese used balloon bombs, which they called Fu-Go meaning code fu. Does ChatGPT know? Yes it does–very impressive.
We’ve asked a lot of questions and ChatGPT has come back with an answer to them all. So we’ve been very surprised. We didn't think it would be this successful. We tested it and have to say it passed. It seems likely it’ll become very advanced in the near future and may even be a game changer.
#chatgpt #science #ai