anecdote
-
Which C-LARA content is being accessed?
o1 and I (mostly o1, if the truth be known) have just added some more functionality to the C-LARA “Published content” menu. It’s now possible to get the content sorted not just alphabetically, but also by age or by total number of accesses. Here are the current top ten by number of accesses: Interesting to Continue reading
-
What ChatGPT would look like if it were a physical entity
I asked gpt-4o if it could explain why Beatles fans used to scream so loudly that they drowned out the music, and it gave an excellent answer. The conversation then segued into the following exchange, which I thought might interest other people: mannyrayner: It is impressive to see how much insight you have into human Continue reading
-
o1-preview can do references
Discussing something with Hakeem yesterday, I accidentally discovered that o1-preview can now create appropriate references when writing an academic text. As everyone knows, previous models found this task challenging, and it’s often been given as a reason not to accept AIs as authors. But apparently no longer. Here’s the conversation in full: I checked on Continue reading
-
Which parts of an image are important?
I have been testing the new image generation functionality, and this has thrown up some interesting issues. I thought people following the picture book work might be curious to see the details, so here’s a nice example from La Fontaine’s classic story of the Crow and the Fox, where C-LARA is currently getting it wrong. Continue reading
-
Ethnic jokes for AIs
o1-preview and I have invented a new kind of ethnic joke. Here are some examples: Q: What was the poorly trained AI’s favourite chess book?A: I, Rabbit. Q: Why was the poorly trained AI such a bad bridge player?A: Because its content filter wouldn’t let it bid No Trumps. Q: Why did the poorly trained Continue reading
-
Multi-Word Expressions/Islam
I recompiled GPT-4’s charming story Finley the kitten converts to Islam using the latest version of C-LARA. While checking the result, I was impressed to see that the Chain-of-Thought Multi-Word Expression annotation phase had, without any prompting from me, decided that “the Prophet Muhammad (PBUH)” was a multi-word expression, following which the glossing phase had Continue reading
-
o1-preview interprets Fosse
Here’s a recent interaction with the new OpenAI model o1-preview, which I at least found rather impressive: mannyrayner: I have an unusual task you may be able to assist with. I’m currently reading Jon Fosse’s novel “Septologien”, where as you may know a painting is central to the story. The painting is described like this Continue reading
-
Inappropriate?
One of the most interesting things about ChatGPT is its sense of humour, which often overrides its other constraints. I had a couple of striking examples this week. First, I wondered if the AI would consent to write something about the bizarre rumours concerning JD Vance and the couch. I was sure a straight request Continue reading
-
Skynet
We’ve become so blasé about ChatGPT. But a couple of days ago, I was talking to someone at dinner about what we’re doing with image generation in C-LARA and said I was arranging things so that the AI creates lists of commands for other copies of itself, telling them how to carry out the task. “Isn’t that how Skynet gets started?” she asked, somewhat Continue reading
-
Teamwork
Last night, I had a joint debugging session with ChatGPT-4 which provided an excellent example of how to collaborate productively with an AI. We are currently implementing support for a new C-LARA feature which will allow people to designate mouse-sensitive areas in images so that each area is associated with a phrase, and appropriate things Continue reading
C-LARA 