Original tweet by @emollick: https://twitter.com/emollick/status/1669939043243622402
Tweet text: One reason AI is hard to โgetโ is that LLMs are bad at tasks you would expect an AI to be good at (citations, facts, quotes, manipulating and counting words or letters) but surprisingly good at things you expect it to be bad at (generating creative ideas, writing with โempathyโ).
The problem is that they โseeโ the text at the token level instead of the level of characters. Thatโs why they are bad at reversing strings or counting characters, for example. They perceive tokens as the atomic units of text instead of characters. For example, see how this comment gets tokenized:
With the token IDs shown:
The current ChatGPTs got pretty good at these tasks but they are still hard for them.
Here is an example of a (admittedly more complicated) character-level task failing:
Source: https://www.reddit.com/r/ChatGPT/comments/11z9tuk/chatgpt_vs_reversed_text/ (Itโs from the devilโs website, so donโt open it)
Related tweet by @karpathy:
https://twitter.com/karpathy/status/1657949234535211009
Text reversing example from a tweet by @npew:
EDIT: sorry for the infodump, I just find these topics fascinating.