Writing Puzzles with OpenAI o1

[Music] so I want to talk about a prompt that GPT forl really struggles with but our new model 01 preview can do pretty well and the prompt is simple it's write a six-line poem about squirrels playing koalas at soccer that meets the following constraints in line two the last word should end with i in line three the second word begins with u in line five the second to last word is eucalyptus and in the final line each word has two syllables so first we'll try with GPT 40 and we'll see that the answer from GPT

40 meets some of the constraints but not all of them the reason it's hard for GPT 40 is it has to get it correct on the first try it can't check that it meets the constraints and then revise the poem now let's try the same poem with o1 preview and we'll see that differing from gb24 o1 preview starts thinking before giving the final answer and you can view a summary of the thinking process of the model so first you could see it's starting to think about uh different words for rhyming then you could see it

wants to make sure the last word matches I it thinks about words like Alibi uh it's analyzing word endings and it's thinking about uh words like ski um then it's piecing together phrases uh but it thinks they don't quite fit it's thinking about phrases where the second word starts with U then it's tweaking the words to fit the two syllable rule for line six um it's digging into various two- syllable word combinations um then it's checking whether the poem aligns with all the guidelines um it's working through the poem to analyze the soccer aspect and

now let's look at the final poem so in the secondine line the word safari does end with i in the third line uh the second word unleash does begin with you in the second to last line um eucalyptus uh uh the second to last word is eucalyptus and finally in the final line under Moonlight creature scatter uh indeed each word has two syllables so this is an example of a prompt where because the model can uh at um uh candidates and do reasoning before giving the final answer it's able to give a higher quality response