4 ms·

Eh. I predict that little will change in the broader tech economy because of AI in 2023. Over the next 3-5 years, yes, but not the next 12 months.

rmason 4y ago
We are at the inflection point with AI. The next version of ChatGPT to be introduced in May 2023 will be 10X as the current version. Then the May 2024 version will be another 10X or 100X the current version. We may need another law such as Moore's law for chips for this to register with people. I humbly suggest that it be called Altman's law.
- pjscott 4y ago
  I saw a copy of "Artificial Intelligence: A Modern Approach" on the shelf yesterday and couldn't help wondering if there would ever be another edition. It has some great, foundational stuff in it -- but the techniques described are about as far from the current state of the art as a Byzantine dromon is from an aircraft carrier. Russell and Norvig would pretty much have to rewrite the book from scratch, and by the time the draft was ready for editing their approach would probably no longer be modern. I'm calling it now: if AI isn't the biggest news of the next ten years, then the reason will probably be because it got upstaged by World War 3 or something similarly dramatic. And it baffles me how little press the subject gets.
- thinkingkong 4y ago
  What are you basing this guess on? What does “10x better ai” even mean?
  - fckgnad 4y ago
    It's an arbitrary number with no exact meaning. Clearly, that number is not based off of any real metric. He is just ironically using quantities to qualitatively express the speed of progress.
  - imtringued 4y ago
    If AI progress was linearly measurable we would have achieved AGI by now.
- omnicognate 4y ago
  On what basis do you say this? Is it anything more than your gut feeling? How much more training data do you think will be needed to bring about this 10x or 100x improvement? Do you have any expectations as to how it will be obtained? Also, what's being multiplied by 10 or 100 in this scenario?
- fckgnad 4y ago
  It's in theory possible for the technologically to grow this quickly. However, there will be a delay in human adoption because of how large of a paradigm shift it is. Artists are already using copyright law to attack AI. Once other parts of our economy are hit you could see more vicious attacks and alternative strategies used against AI. It's possible for even laws used to limit AI in the same way tariffs work or we might even see resistance similar to the Butlarian jihad. Who knows? That's all hypothetical though... Even if none of these resistance scenarios arise, I still think adoption will not happen as fast as technological development.
fckgnad 4y ago
I actually agree with this and i never said anything against it. The next year is still extremely important. Here's a better way for me to put it: chatGPT and stable diffusion didn't change anything about the economy in 2022 and they won't change anything in 2023. However, these two technologies are still inflections points and that is why they are important despite Zero change. In 2023 we will see much more of these inflection points related to AI. None of these will change the economy in 2023, but they are the first stepping stones of what will change everything in the years to come.

rmason 4y ago

We are at the inflection point with AI. The next version of ChatGPT to be introduced in May 2023 will be 10X as the current version. Then the May 2024 version will be another 10X or 100X the current version. We may need another law such as Moore's law for chips for this to register with people. I humbly suggest that it be called Altman's law.

pjscott 4y ago
I saw a copy of "Artificial Intelligence: A Modern Approach" on the shelf yesterday and couldn't help wondering if there would ever be another edition. It has some great, foundational stuff in it -- but the techniques described are about as far from the current state of the art as a Byzantine dromon is from an aircraft carrier. Russell and Norvig would pretty much have to rewrite the book from scratch, and by the time the draft was ready for editing their approach would probably no longer be modern. I'm calling it now: if AI isn't the biggest news of the next ten years, then the reason will probably be because it got upstaged by World War 3 or something similarly dramatic. And it baffles me how little press the subject gets.
thinkingkong 4y ago
What are you basing this guess on? What does “10x better ai” even mean?
- fckgnad 4y ago
  It's an arbitrary number with no exact meaning. Clearly, that number is not based off of any real metric. He is just ironically using quantities to qualitatively express the speed of progress.
- imtringued 4y ago
  If AI progress was linearly measurable we would have achieved AGI by now.
omnicognate 4y ago
On what basis do you say this? Is it anything more than your gut feeling? How much more training data do you think will be needed to bring about this 10x or 100x improvement? Do you have any expectations as to how it will be obtained? Also, what's being multiplied by 10 or 100 in this scenario?
fckgnad 4y ago
It's in theory possible for the technologically to grow this quickly. However, there will be a delay in human adoption because of how large of a paradigm shift it is. Artists are already using copyright law to attack AI. Once other parts of our economy are hit you could see more vicious attacks and alternative strategies used against AI. It's possible for even laws used to limit AI in the same way tariffs work or we might even see resistance similar to the Butlarian jihad. Who knows? That's all hypothetical though... Even if none of these resistance scenarios arise, I still think adoption will not happen as fast as technological development.

fckgnad 4y ago

I actually agree with this and i never said anything against it. The next year is still extremely important. Here's a better way for me to put it: chatGPT and stable diffusion didn't change anything about the economy in 2022 and they won't change anything in 2023. However, these two technologies are still inflections points and that is why they are important despite Zero change. In 2023 we will see much more of these inflection points related to AI. None of these will change the economy in 2023, but they are the first stepping stones of what will change everything in the years to come.