3 ms·

I can't see an end in sight for the capabilities of these systems anymore. I've lost a LOT of sleep regarding this over the last week as I work on a prototype t

I can't see an end in sight for the capabilities of these systems anymore. I've lost a LOT of sleep regarding this over the last week as I work on a prototype to demo to my org's leadership. In our sector (B2B consulting), this sort of technology is essentially "adapt or die". Additionally, I see this technology as a way to level the playing field for a brief moment. F500 will still be there, but I think the constituents are going to change dramatically. ChatGPT is a warning shot for me. The technology had been there for a while, but this arrangement is particularly compelling (concerning). Even if the current iteration doesn't achieve "criticality", the next one almost certainly will. I am seeing 2 paths for a developer: embrace these tools or become the next generation of luddite. I've got no clue if this is going to eliminate jobs, but I would certainly prefer to be on the vendor-side of this software than the receiving end. The roles inside of a typical org chart are going to change very quickly - That much is certain to me.

isaacremuant 4y ago
If it's good. We can leverage it, if it's not, what's the problem? That medium bar people or systems will rely on it and produce medium bar software? There's already a lot of that through outsourcing and bad decisions and incentives. Right now it's definitely not that good to be a real aid to a software developer who knows his stuff but will probably help brainstorm when something is obscure, badly documented or unclear. People who don't know the subject matter will be impressed because they won't be able to judge the output but those who do know, will not be so impressed. It exposes a lot of the unnecessary things (cover letters and corporate drivel) and that's laudable. It's still a text in text out thing that can only do what it's trained on and not all the other sci-fi things people seem to think it is. There's no need to overreact, even if definitely people without understanding of the tech and crazy for the latest buzzword will.
- bob1029 4y ago
  > There's no need to overreact, even if definitely people without understanding of the tech and crazy for the latest buzzword will. Agreed. In my particular case, I looked at this last year and dismissed it as a shiny distraction. Looking at it again with a fresh perspective, I have arrived at a different conclusion. The part that bothers me is not replacing the ivory tower of engineering or consulting with a robot. It's the other 99% of the market that can be captured virtually overnight.
  - Kye 4y ago
    I know I keep hammering on Bing Chat lately and I feel like a shill, but: it's really, really good. Bing already handled the stuff Google is still good at well enough, and now it's better at the stuff Google used to be good at. Two Goliath companies forced to compete in the blue ocean of practical LLMs. It's going to be a wild ride of innovation and cool new stuff. When was the last time Google faced real competition? This is it. This is the next era of tech. Hopefully someone decides to compete on consent, credit, and compensation for the people whose work these models depend on. Google's bargain was: We index your stuff to make it searchable, and we run ads on it while sending traffic to you. The LLM chat-based search bargain is similar: we scoop up your work and help you make more. The glitch is that LLMs have the potential to negate the you and your and take over, like Google has with that bargain in trying to answer all queries on-search instead of sending traffic. Someone has to solve the problem of people needing to work for compensation to survive. I really can take or leave copyright and IP in general. It exists to get people paid for creating stuff we enjoy and benefit from as a society. I would happily hand it all over to these machines to help me come up with new stuff if I weren't struggling to survive when my only real skills are at risk of being turned into a mush of data that people will acclimate to and forget they ever cared about craft or the people who used to make it. Wouldn't it be wild if our discussions about this became input for the models, and those models came up with some good solutions?
cj 4y ago
I'm more 50/50 on it than you. I have a feeling it will take a few years at a minimum to see reliable, trustworthy, accurate, useful ways of using it. But I do agree it's adapt or die (if not now, eventually)

isaacremuant 4y ago

If it's good. We can leverage it, if it's not, what's the problem? That medium bar people or systems will rely on it and produce medium bar software? There's already a lot of that through outsourcing and bad decisions and incentives. Right now it's definitely not that good to be a real aid to a software developer who knows his stuff but will probably help brainstorm when something is obscure, badly documented or unclear. People who don't know the subject matter will be impressed because they won't be able to judge the output but those who do know, will not be so impressed. It exposes a lot of the unnecessary things (cover letters and corporate drivel) and that's laudable. It's still a text in text out thing that can only do what it's trained on and not all the other sci-fi things people seem to think it is. There's no need to overreact, even if definitely people without understanding of the tech and crazy for the latest buzzword will.

bob1029 4y ago
> There's no need to overreact, even if definitely people without understanding of the tech and crazy for the latest buzzword will. Agreed. In my particular case, I looked at this last year and dismissed it as a shiny distraction. Looking at it again with a fresh perspective, I have arrived at a different conclusion. The part that bothers me is not replacing the ivory tower of engineering or consulting with a robot. It's the other 99% of the market that can be captured virtually overnight.
- Kye 4y ago
  I know I keep hammering on Bing Chat lately and I feel like a shill, but: it's really, really good. Bing already handled the stuff Google is still good at well enough, and now it's better at the stuff Google used to be good at. Two Goliath companies forced to compete in the blue ocean of practical LLMs. It's going to be a wild ride of innovation and cool new stuff. When was the last time Google faced real competition? This is it. This is the next era of tech. Hopefully someone decides to compete on consent, credit, and compensation for the people whose work these models depend on. Google's bargain was: We index your stuff to make it searchable, and we run ads on it while sending traffic to you. The LLM chat-based search bargain is similar: we scoop up your work and help you make more. The glitch is that LLMs have the potential to negate the you and your and take over, like Google has with that bargain in trying to answer all queries on-search instead of sending traffic. Someone has to solve the problem of people needing to work for compensation to survive. I really can take or leave copyright and IP in general. It exists to get people paid for creating stuff we enjoy and benefit from as a society. I would happily hand it all over to these machines to help me come up with new stuff if I weren't struggling to survive when my only real skills are at risk of being turned into a mush of data that people will acclimate to and forget they ever cared about craft or the people who used to make it. Wouldn't it be wild if our discussions about this became input for the models, and those models came up with some good solutions?

cj 4y ago

I'm more 50/50 on it than you. I have a feeling it will take a few years at a minimum to see reliable, trustworthy, accurate, useful ways of using it. But I do agree it's adapt or die (if not now, eventually)