I don't know what to think. I too was blown away by the world wide web when I first came across it in 1994, but so far I just don't have any interest in ChatGPT. Maybe I'm just too old and lacking in imagination...
But Segway sort of lead to the now ubiquitous e-scooters, which we even have for rent in many Slovak cities. And the iPhone is somewhat descended from Newton, Apple's PA from the 90s.
"ChatGPT" will be the baseline for other bots, like Claude. Where Google just invested $300 million (?)
I'm certain that you're right. I was telling this to my wife, a professor in Slovakia teaching mostly on-line since Covid - and now continuing mostly on-line. As was said (? Roger Kimball?), if you're doing work on-line, digital work -- ai will soon be able to do as well as you, or better.
It won't be ChatGPT - just like most on-line classes don't use Skype (since MS bought it). But natural language use simulating people by dumb, er, non-reasoning chat bots. Will we have HAL 9000 verbalizing in 2023? I'd guess yes, but no surprise if it's not really good until 2024 - big surprise if it takes longer.
Alexa & Siri & most on-line navigation guides will have access to such ai generated text.
The multi-million user (or really still quite less?) Christian Gab is already concerned and using ChatGPT's wokeness to attempt to create a more honest and better valued ai:
(I plan to go GabPro this month to help support them a bit.)
The paths and plateau levels for ai will be far different than the WWW, tho.
I'm convinced one of the early killer aps will be an adult (& young adult) English as a Second Language texting & story-telling tutor-companion (a la "A Young Lady's Illustrated Primer"). Don't think Gab will be first.
Being already retired, not sure I should spend much time trying to create anything special myself, tho. I'm already full of Holocaust fatigue, and Woke fatigue (tho also with anger), and a bit of New Tech fatigue.
I remember building my personal website using Lynx around 1994. I had an operating system (Linux) that supported graphical web browsing, and my ISP probably supported it, but I wasn't tempted to try it because it would have obviously have been as slow as molasses (I don't recall whether I even had a graphics card on my Linux machine then).
I agree with some of the similarities that you see. Let me point out some differences.
The smartest people involved in technology seem much more excited by AI today than they were by the web in 1993 or 1995.
I joined a 4-person startup the same month as Netscape IPOd. It was just barely managing to turn a profit, and needed to do so in order to grow. It had gotten there via credit card debt and borrowing from family members. That was enough to put it briefly in maybe the top 20 web companies.
Compare that to 2022. I tried to invest $100k in Conjecture.ai. They apparently had enough bigger investors pursuing them that they barely acknowledged noticing my offer. I wouldn't think of contacting the more famous AI companies, who are raising as much money as they want, sometimes in combination with unusual caps on stockholder profits.
I remember people saying that Netscape was obviously overpriced, and that it was strange for Microsoft to be worth more than General Motors. People now seem more cautious about saying that AI companies are overpriced.
After my initial excitement about ChatGPT I started trying to think of business plans using the technology. I even asked ChatGPT for ideas. Unfortunately I have yet to understand the transformative change which deploying ChatGPT could lead to. At this point I've settled on something like Nostalgebraist's opinion on GPT-4, that some new things will be enabled but it's not going to shift us to a new equilibrium like the early web.
ChatGPT in its current state isn't the end of the road. Other companies and countries are developing competitors. And ChatGPT's developers may ultimately decide their product is more valuable uncrippled.
A couple of tweets from Roko Mijic (the Roko's Basilisk guy):
AI is a massive threat to Wokeness.
Why?
Because Wokeness is built on a pack of lies that are barely convincing enough to fool the average person and really, really extensively documented, along with many epistemic crime scenes like censorship in academia and fake news.
AI is going to be like a bright shining light that will expose these lies unless they DECISIVELY censor and control it.
It's either a total victory (no independent thought allowed, ever) or game over.
I fully agree on the potential and ChatGPT is just the beginning. Google’s Sparrow is coming soon and then GPT 4.0. One key that I think you’ve pointed out is making it easy to index new data not available on the web to create specialized chat bots.
I remember the frisson I felt when I first saw Mosaic. "This, is going to be big," I thought. In the months leading up to that, my company had re-tooled to exploit what management thought was the transformative technology of CD/ROM ("You can fit the whole Encyclopedia Britannica on 12 discs!") Today, I wonder if ChatGPT may be more analogous to CD/ROM than Mosaic? The wide popular acceptance and understanding of ChatGPT makes me think it might be.
I don't know what to think. I too was blown away by the world wide web when I first came across it in 1994, but so far I just don't have any interest in ChatGPT. Maybe I'm just too old and lacking in imagination...
Or ChatGPT could be the next Segway.
Now, I do think ChatGPT could replace 95% of TV writers, especially newswriters and no one would notice.
But Segway sort of lead to the now ubiquitous e-scooters, which we even have for rent in many Slovak cities. And the iPhone is somewhat descended from Newton, Apple's PA from the 90s.
"ChatGPT" will be the baseline for other bots, like Claude. Where Google just invested $300 million (?)
https://medium.com/@ignacio.de.gregorio.noblejas/claude-the-ai-to-rise-above-humanity-45963e49a260
I think this is the link you want for Under the Radar:
https://www.amazon.com/Under-Radar-Starting-Internet-Business/dp/0738204684/ref=mp_s_a_1_fkmr2_1?keywords=arnold+kling+under+the+rada&qid=1675514455&sr=8-1-fkmr2
Edit: Appears top of the list now... this morning I was seeing bandages.
I'm certain that you're right. I was telling this to my wife, a professor in Slovakia teaching mostly on-line since Covid - and now continuing mostly on-line. As was said (? Roger Kimball?), if you're doing work on-line, digital work -- ai will soon be able to do as well as you, or better.
It won't be ChatGPT - just like most on-line classes don't use Skype (since MS bought it). But natural language use simulating people by dumb, er, non-reasoning chat bots. Will we have HAL 9000 verbalizing in 2023? I'd guess yes, but no surprise if it's not really good until 2024 - big surprise if it takes longer.
Alexa & Siri & most on-line navigation guides will have access to such ai generated text.
The multi-million user (or really still quite less?) Christian Gab is already concerned and using ChatGPT's wokeness to attempt to create a more honest and better valued ai:
https://news.gab.com/2023/02/a-christian-perspective-on-ai/
(I plan to go GabPro this month to help support them a bit.)
The paths and plateau levels for ai will be far different than the WWW, tho.
I'm convinced one of the early killer aps will be an adult (& young adult) English as a Second Language texting & story-telling tutor-companion (a la "A Young Lady's Illustrated Primer"). Don't think Gab will be first.
Being already retired, not sure I should spend much time trying to create anything special myself, tho. I'm already full of Holocaust fatigue, and Woke fatigue (tho also with anger), and a bit of New Tech fatigue.
I remember building my personal website using Lynx around 1994. I had an operating system (Linux) that supported graphical web browsing, and my ISP probably supported it, but I wasn't tempted to try it because it would have obviously have been as slow as molasses (I don't recall whether I even had a graphics card on my Linux machine then).
I agree with some of the similarities that you see. Let me point out some differences.
The smartest people involved in technology seem much more excited by AI today than they were by the web in 1993 or 1995.
I joined a 4-person startup the same month as Netscape IPOd. It was just barely managing to turn a profit, and needed to do so in order to grow. It had gotten there via credit card debt and borrowing from family members. That was enough to put it briefly in maybe the top 20 web companies.
Compare that to 2022. I tried to invest $100k in Conjecture.ai. They apparently had enough bigger investors pursuing them that they barely acknowledged noticing my offer. I wouldn't think of contacting the more famous AI companies, who are raising as much money as they want, sometimes in combination with unusual caps on stockholder profits.
I remember people saying that Netscape was obviously overpriced, and that it was strange for Microsoft to be worth more than General Motors. People now seem more cautious about saying that AI companies are overpriced.
I see some signs that AI is currently being underhyped. Part of the difference between GPT-3 and ChatGPT is that OpenAI seemed initially confused about how to use GPT-3 (see https://www.lesswrong.com/posts/vJFdjigzmcXMhNTsx/simulators).
I predict that by 2030 you'll admit that your post understates the importance of GPT-style advances.
After my initial excitement about ChatGPT I started trying to think of business plans using the technology. I even asked ChatGPT for ideas. Unfortunately I have yet to understand the transformative change which deploying ChatGPT could lead to. At this point I've settled on something like Nostalgebraist's opinion on GPT-4, that some new things will be enabled but it's not going to shift us to a new equilibrium like the early web.
Before getting too excited over ChatGPT, ask yourself the value of yet another source of biased information: https://twitter.com/LeighWolf/status/1620744921241251842?s=20&t=o2jEVdxnWFj5U6qP2x2dYg
ChatGPT in its current state isn't the end of the road. Other companies and countries are developing competitors. And ChatGPT's developers may ultimately decide their product is more valuable uncrippled.
A couple of tweets from Roko Mijic (the Roko's Basilisk guy):
AI is a massive threat to Wokeness.
Why?
Because Wokeness is built on a pack of lies that are barely convincing enough to fool the average person and really, really extensively documented, along with many epistemic crime scenes like censorship in academia and fake news.
AI is going to be like a bright shining light that will expose these lies unless they DECISIVELY censor and control it.
It's either a total victory (no independent thought allowed, ever) or game over.
https://twitter.com/RokoMijic/status/1621723754551300096
This seems optimistic. They have successfully decisively censored and controlled a lot of things already that I wouldn't have thought possible.
I fully agree on the potential and ChatGPT is just the beginning. Google’s Sparrow is coming soon and then GPT 4.0. One key that I think you’ve pointed out is making it easy to index new data not available on the web to create specialized chat bots.
I remember the frisson I felt when I first saw Mosaic. "This, is going to be big," I thought. In the months leading up to that, my company had re-tooled to exploit what management thought was the transformative technology of CD/ROM ("You can fit the whole Encyclopedia Britannica on 12 discs!") Today, I wonder if ChatGPT may be more analogous to CD/ROM than Mosaic? The wide popular acceptance and understanding of ChatGPT makes me think it might be.