Gresham College
Could AI replace stand-up comedians and scriptwriters? This may not be an impossible dream if you accept that nothing we do is forever beyond the scope of computer modelling.
This lecture explores attempts to create jokes from rules, and programs that create not-quite-relevant responses that hearers can make meaningful and comic. Will computers ever tell good jokes?
A lecture by Yorick Wilks
The transcript and downloadable versions of the lecture are available from the Gresham College website:
http://www.gresham.ac.uk/lectures-and-events/ai-humour
Gresham College has been giving free public lectures since 1597. This tradition continues today with all of our five or so public lectures a week being made available for free download from our website. There are currently over 2,000 lectures free to access or download from the website.
Website: https://gresham.ac.uk
Twitter: https://twitter.com/greshamcollege
Facebook: https://facebook.com/greshamcollege
Instagram: https://instagram.com/greshamcollege
Would a machine with AI ever get to appreciate the humour or the wit in the following joke?
There are two snowmen standing in a field when one turns to the other to ask ' Can you smell carrots?'
The first joke is funny, but you would have to be a bit of a fool to believe a machine actually wrote it.
Puns shouldn't be too difficult for AI to come up with.
The best level of humour that AI can hope to obtain within the next ten to twenty years would be that which is typical found in Christmas crackers. These jokes are often founded upon feeble puns and riddles. This suggests that coded knowledge will ultimately win out.
The New Yorker cartoon (although very weak, as many New Yorker cartoons were and are) can easily be converted into a verbal joke (no weaker IMO) as follows: A woman sits down for an interview. The interviewer, looking straight at her, says: 'I see from your resume you are a woman'.
Proud to have the last name of Gresham.
I am of the opinion that the statistical machine learning approach will win in the end for not just humor, but for all problems that seem intractable. Consider already the progress made by OpenAi with both GPT3 and Dall-E. A classical, rules based approach to these problems would never be possible for the simple reason that humans are incapable of reasoning about rules above a certain level of complexity.
Laughing jack is going to haunt me for the rest of my life. The joke is that it can't be unseen. Ample training data for the curious little bird hovering over my shoulder.