Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I’m holding out for Open AI Co Pilot!
Screenshot 2024-12-06 at 11.35.40.png
 
I get that its aimed at researchers and industry rather than the everyperson but I can't help think that scientists would earn more kudos for asking for the charitable use of modern games consoles for another 'Folding@Home' initiative. What they could do with todays machines and internet speeds is unthinkable!
 
  • Like
Reactions: NoGood@Usernames
The arguement would be that if GPT can be used to cheat on an assessment then the assessment methods need to be changed.

Completely agree with that.

To be fair, I'm on the mathematics side of things, and it's pretty much impossible to use ChatGPT to cheat. It does an absolutely terrible job of any remotely abstract or even procedural problem. Perhaps a 10% hit rate on a cohesive, well structured and usable answer for an exam. The other 90% of the time it confidently gives you a completely wrong answer or process and does so in a plainly identifiable way. So for scenarios where you have to provide a fully written answer or proof, it'll be obvious. And for scenarios where you have to provide a simple answer, say computer marked tests, then it'll confidently be wrong.

End game is only **** students and cheaters get eliminated, which is the point of the examination process.

Note I'd expect 80%+ score on a mathematics test to be a credible demonstration of someone actually knowing the material.
 
Great move! This makes $20/month look super cheap in comparison. I recommend a $2000 tier as well, to make $200 look like a good deal! 🤣

(ok, seriously - for those who need these features or don't want to worry about the advanced voice mode limit, it's good to have an option like this - it's clearly not for everyone).
 
  • Haha
Reactions: lazyrighteye
Completely agree with that.

To be fair, I'm on the mathematics side of things, and it's pretty much impossible to use ChatGPT to cheat. It does an absolutely terrible job of any remotely abstract or even procedural problem. Perhaps a 10% hit rate on a cohesive, well structured and usable answer for an exam. The other 90% of the time it confidently gives you a completely wrong answer or process and does so in a plainly identifiable way. So for scenarios where you have to provide a fully written answer or proof, it'll be obvious. And for scenarios where you have to provide a simple answer, say computer marked tests, then it'll confidently be wrong.

End game is only **** students and cheaters get eliminated, which is the point of the examination process.

Note I'd expect 80%+ score on a mathematics test to be a credible demonstration of someone actually knowing the material.
I did read on the old socials about a maths teacher who fed a mark scheme into GPT and uses it to grade all his tests!
 
All the comments indicate to me that nobody here is using what they offer for what it’s intended to be for.
And for those who use it for the purpose it’s made for, 200$ is really not a lot of money at all and you can’t replicate this with less costs at “home” or in your own “IT dungeon” at the office. It’s just not possible.

Spot on... For a professional office/organization (patent attorneys and other law offices, bio-tech sciences, various research disciplines, universities, architects, defense strategists, etc) $200/mo is a tiny overhead cost and likely offers outstanding value for time saved and results produced.

It's not for the average futzing-around-at-home computer user wanting to play around with the latest tech.
 
I wonder what ChatGPT Pro thinks about the price :).

Here's what regular ChatGPT thinks:

"Whether ChatGPT Pro is worth $200 per month depends on your needs and how you intend to use it. If you're using AI extensively for professional purposes (e.g., for coding, writing, research, or content generation), the $200 per month could be a good investment. For casual users or those primarily using AI for personal entertainment or basic queries, the free version might be sufficient."
 
pure greed....thats all this is....there is no way that the current version of AI is even worth half that....
Then get the subscription that is a tenth of the price? It's impossible for any of us to know what this subscription is worth if we're not going to use it for the purposes it was intended anyway.
 
  • Like
Reactions: citysnaps
Ollama and Open Web UI do 90% of whatever these online versions do, but privately and with a choice of new models every week. For that $2400 a year OpenAI wants you can get a top end new Mac or beefy GPU for a PC every two years.
 
Yeah....
I mean it would be like a human, reading books, watching video's and going to lessons and learning to be clever based upon other people's previous work.
And then, after a few years of them learning, expecting people to pay them for what they are a person can offer.

Those greedy humans who studied for years on other people's work, should offer their services for free ;)
Terrible analogy. I guess you have never generated meaningful content and wanted to be compensated for your work?

Every book read… was purchased from the publisher/author (even books in the library (taxpayer dollars))

Every TV show watched… paid for with subscription or advertising dollars

Every news site…

Every movie…

Every university course…

Every YouTube video….

Copyright exists for a reason. OpenAI can use these data sources. They need to pay/ask for permission like everyone else. They haven’t. They didn’t ask.
 
Terrible analogy. I guess you have never generated meaningful content and wanted to be compensated for your work?

Every book read… was purchased from the publisher/author (even books in the library (taxpayer dollars))

Every TV show watched… paid for with subscription or advertising dollars

Every news site…

Every movie…

Every university course…

Every YouTube video….

Copyright exists for a reason. OpenAI can use these data sources. They need to pay/ask for permission like everyone else. They haven’t. They didn’t ask.

Copyright also expires for a reason otherwise a handful of corporates and rich families would own every form of creative outlet. They would own the phrase ‘I love you’ and sue you if you ever used it professionally.

There’s nothing wrong with learning from pre-existing material and creating something newish or fresh from it, whether a person does it or if deep learning assists us to do it faster. There are fair use laws that allow us that and if we didn’t learn or copy from those before us we would learn and create nothing. We would be a mentally dead species. Steve Jobs himself said great artists steal (he did not invent this phrase).

Copyright is relatively new in the timeline of human history. For the majority of our existence we were always copying and studying each other and it was encouraged by the great masters of the arts. The masters even took credit for the work their apprentices were doing. From Da Vinci down to Warhol they sat around taking credit while young artists did the work.

As for the music industry, musicians steal, copy, sample and regurgitate all the time. Most pop music is exactly that. Sometimes they blatantly lie about it and because of their corporate power behind them they don’t give a damn.

A lot of deep learning is open source and free. Use it if you want without fear, but at the same time it’s not some dangerous voodoo magic that can achieve anything. It’s actually really boring once you get used to it and see how janky it is.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.