Hacker News new | past | comments | ask | show | jobs | submit login
The Future of Employment: How Susceptible Are Jobs to Computerisation? (2013) [pdf] (ox.ac.uk)
10 points by JSeymourATL on Jan 30, 2023 | hide | past | favorite | 13 comments



From the report, "Our findings thus imply that as technology races ahead, low-skill workers will reallocate to tasks that are non-susceptible to computerisation –i.e., tasks requiring creative and social intelligence. For workers to win the race, however, they will have to acquire creative and social skills."

Interesting, if this report was written today it would have a different conclusion.

Bottom line, you can't predict the future no matter how well researched and cited your report is.


Even at the time , there was the genetic algorithm competition , gecco , which showed computers could be creative.


True, also now that I think about it, the conclusion was flawed even without gecco. It implied that creative work is low skill by default. My observation has been that acceptable creative work has been the result of lots of experience and hard work. You can't just decide to be creative with a few months of work. It takes time, hard work, and patience to be successful.


You can't predict the future by extrapolating past trends. That's kinda what "well researched" means, especially if it wasn't controversial, which it probably isn't if it was widely cited.


> it would have a different conclusion.

what would today's conclusion be?


"shut up you luddite, it's perfectly acceptable for your artist job to be automated away, progress is inevitable and all of your objections about how this is playing fast and loose with copyright are meaningless" is pretty much what I get from a lot of Hacker News when I talk about how AI art is not making me happy for the future prospects of my field.


I think these artists should be able to prohibit their work being used in a training set

Somewhat related - how does copyright work on the output of these AIs? Can a piece of code be the copyright owner of a picture?


> how does copyright work on the output of these AIs?

the person who pressed the button to start the AI generation for that particular output _should_ own the copyright to it (and also be liable for the infringement if it does).


> Can a piece of code be the copyright owner of a picture?

No. Copyright requires a human creator.


AI has generally been shown to have high creativity (e.g. Dall-E). However, AI is still lagging in other functional areas where humans excel (e.g. driving).

So, the conclusion might be the actual opposite that AI will first achieve human level creativity before (if ever) acquiring human functional skills.


Creativity requires solving unexplored problems. Painting a picture isn't creative, nor is it particularly marketable.

Figuring out a way to use your love of panting in order to provide a marketable product or service requires creativity, but I'm not sure Dall-E succeeds on that front. What marketable service it is able to provide has been left to the creativity of the humans who operate OpenAI.


Today's humans should train for applications where they have a lasting advantage over machines: physical-object-manipulation tasks of low power and low-to-moderate repetitiveness, where the humans' cheap and nimble actuators outperform machinery on a cost basis.


That brings an interesting question, what is the fully-loaded cost of these nimble actuators. It's not like they appear from the womb ready to be a worker




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: