Sigh. I am going to spend too much of the rest of my life squinting at glib, confident prose and trying to figure out whether it makes any sense at all. “Your scientists were so preoccupied with whether or not they could, they didn't stop to think if they should.”
Looking forward to Scientist GPT. Did they fudge the data? Nope! They pulled it straight out of their model. (Someone needs to make a "publish paper" button backed by Chat GPT)
It's that intersection of confident and so on its face, even at first glance, wildly incorrect that it just makes you feel like you're the stupid one who's missing something obvious
Right? I have spent a long time building up an ability to measure quality of thought via quality of prose as a proxy. This breaks that almost entirely, because they decided to train it for glibness with little regard for correctness.
Actually, so far I've seen humanities people discard chatGPT output faster than sciences.
In the humanties there is lots of well-written text of no useful depth, so academics in the humanities get good at filtering it quickly. In the sciences, it seems "this is well written and well structured" is treated as a filter to assume the text probably contains something worth reading -- of course eventually it is found to be nonsense.
> ...a professor in the U.K., used GPT-3, a large language model from OpenAI that automatically generates text from a prompt, to write it. (The whole essay, which Sharples considered graduate-level, is available, complete with references, here.) Personally, I lean toward a B+. The passage reads like filler, but so do most student essays.
> It's difficult to say what these numbers might represent without more context. They could be a list of numerical data, possibly related to a particular topic or theme. They could also be codes or identifiers for something, or possibly even a series of random numbers. Without more information, it's hard to provide a more specific suggestion.
I only gave it the info that was available in the question, all it came up with were different variations of "no idea".
I've seen a lot of similar confident but utterly nonsensical responses from ChatGPT. It really comes across as the ultimate con artist. Project confidence and drown people in so many words that they start doubting themselves instead of ChatGPT. We may have taught AI to gaslight us.
I tried it on the you.com chatbot, which has access to search results of your query. Even though it was already spoiled by receiving a link to this thread in which the correct answer is literally listed, it still managed to come up with nonsense:
> The numbers appear to be arranged in an exponential pattern [1], where each number is approximately 2.3 times larger than the previous number. This pattern can be expressed as: n = 2.3^x, where n is the number and x is the index of the number. So for the numbers you provided, the pattern would be: 925 = 2.3^0 8642 = 2.3^2 37654 = 2.3^3 627418 = 2.3^4
