Creatíon? yes. Quality? maybe. Art? debatable. Accurate information? highly dubious.
I would say 90% of the time I've asked ChatGPT about something, it has given incorrect answers. Most of the time I feel like it's getting more out of me (improving the model) than I get out of it. Example:
Well, I say "has given incorrect answers", but that's not really what it has done as it doesn't 'understand' anything and has no real grasp of facts - it's essentially an extremely fancy version of autocomplete. Impressive, but still ultimately dumb. On the other hand, is this a valid criticism? Do we really understand what it means to 'understand'? Could us humans just be pattern matching too, but just at a much more sophisticated level?
This tweet also makes a very good point:
Paul Graham
@paulg
If AI saves people from having to write, it will also save them from having the ideas that writing engenders.
01:37 AM - 05 Jan 2023
I think generative AI can and will continue to boost human creation, but has a worrying potential to impoverish true human creativity.
Accessibility Specialist. I focus on ensuring content created, events held and company assets are as accessible as possible, for as many people as possible.
Haha Chat GPT and logic puzzles...even simple ones...is so funny. But I also love how it shows us what it "understands" about language and what it doesn't!
Accessibility Specialist. I focus on ensuring content created, events held and company assets are as accessible as possible, for as many people as possible.
Oh I completely agree, Chat GPT 3.5 is something to marvel at in some ways, but those who say they are using it every day to do their work for them...they are like the web3 hype train people, evangelical without objectivity.
And the fact we are training the model, nothing new there! Every single time we do a Google Captcha identifying cars (or words in the past) we have been training AI models.
For me, I am hopeful that the effort we are putting in training AI and into trying to remove AI "hallucinations" makes them more dependable and accurate in the future.
I suppose the question is "is AI more dependable than your average human for answers" and the scary thing is that for most subjects, that answer is probably yes. The second question is "should we depend on AI for answers" and for that the answer, as of today, should almost certainly be no!
And the tweet you shared...it is almost like a prophecy to me! For example I no longer have a sense of direction due to GPS and SatNavs, will I no longer have writing skills if I start using AI to write for me? Quite possibly.
More importantly does that mean that writing as a skill will become less valuable, and instead the ability to fact check and correct errors becomes the new role of writers...will everyone become an editor and a curator? This is where my brain power runs out and I realise I am not smart enough to answer these things lol! 💗
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
Creatíon? yes. Quality? maybe. Art? debatable. Accurate information? highly dubious.
I would say 90% of the time I've asked ChatGPT about something, it has given incorrect answers. Most of the time I feel like it's getting more out of me (improving the model) than I get out of it. Example:
Well, I say "has given incorrect answers", but that's not really what it has done as it doesn't 'understand' anything and has no real grasp of facts - it's essentially an extremely fancy version of autocomplete. Impressive, but still ultimately dumb. On the other hand, is this a valid criticism? Do we really understand what it means to 'understand'? Could us humans just be pattern matching too, but just at a much more sophisticated level?
This tweet also makes a very good point:
I think generative AI can and will continue to boost human creation, but has a worrying potential to impoverish true human creativity.
Another gem I saw today:
Haha Chat GPT and logic puzzles...even simple ones...is so funny. But I also love how it shows us what it "understands" about language and what it doesn't!
Oh I completely agree, Chat GPT 3.5 is something to marvel at in some ways, but those who say they are using it every day to do their work for them...they are like the web3 hype train people, evangelical without objectivity.
And the fact we are training the model, nothing new there! Every single time we do a Google Captcha identifying cars (or words in the past) we have been training AI models.
For me, I am hopeful that the effort we are putting in training AI and into trying to remove AI "hallucinations" makes them more dependable and accurate in the future.
I suppose the question is "is AI more dependable than your average human for answers" and the scary thing is that for most subjects, that answer is probably yes. The second question is "should we depend on AI for answers" and for that the answer, as of today, should almost certainly be no!
And the tweet you shared...it is almost like a prophecy to me! For example I no longer have a sense of direction due to GPS and SatNavs, will I no longer have writing skills if I start using AI to write for me? Quite possibly.
More importantly does that mean that writing as a skill will become less valuable, and instead the ability to fact check and correct errors becomes the new role of writers...will everyone become an editor and a curator? This is where my brain power runs out and I realise I am not smart enough to answer these things lol! 💗