Nat.dev allows you to use their API to run GPT4 without limit, you just pay a certain amount per message (you can start with 5$). A great way to try out GPT4 for yourself with a small investment. I haven't played around with it much yet though but it does keep memory as far as I am aware.
AutoGPT is extremely promising and can't wait for my API as well to get to try it, it should make programming a fair bit easier than it is right now.
The Following User Says Thank You to Firebot For This Useful Post:
The great thing with GPT is not search, but everything else it can do. If you had a personal assistant who messed up when asked what they knew about your grandfather but could still perform all the actual work required of them, like processing documents, writing plans, building websites, coordinating activities etc., and could do that for a salary of $20/month, I would think you'd be pretty pleased.
I for one can't wait to eliminate tons of jobs. Personally I want nothing to do with AI especially AI art. The devaluing of skills is going to be very interesting.
I for one can't wait to eliminate tons of jobs. Personally I want nothing to do with AI especially AI art. The devaluing of skills is going to be very interesting.
I can't control the weather, I just choose how I want to dress for it.
A more optimistic outlook you might adapt is thinking of how much more you (or anyone with access) can do to create value or bring good into the world with these tools than is otherwise possible.
__________________
"If stupidity got us into this mess, then why can't it get us out?"
The great thing with GPT is not search, but everything else it can do. If you had a personal assistant who messed up when asked what they knew about your grandfather but could still perform all the actual work required of them, like processing documents, writing plans, building websites, coordinating activities etc., and could do that for a salary of $20/month, I would think you'd be pretty pleased.
As long as you know which tasks it can be trusted with and/or it's work that you will be validating. I like using it for writing random code snippets because I need to test regardless who wrote the code. I don't know if it would write better or more secure code that a real dev but it's better than me.
Most people aren't going to think about this though and just take whatever it spews out as gospel. There needs to be more work to either validate results or know when to not present them as facts. Would you hire an assistant who often just made stuff up but sounded good?
The Following 2 Users Say Thank You to edslunch For This Useful Post:
I can't control the weather, I just choose how I want to dress for it.
A more optimistic outlook you might adapt is thinking of how much more you (or anyone with access) can do to create value or bring good into the world with these tools than is otherwise possible.
Oh don't get me wrong I wasn't really pointing the finger at you, just a general statement I guess.
Yeah I just don't agree. In the case of art, if any one can create it with zero skill it has no value on top of it being mostly stolen from talented artists. As far as other industries, displacing people doesn't bring any good into the world. They move to industries less affected by AI and drive the value down there. I see a lot of AI as a tool to devalue skills and make money for those at the top.
All the dump trick drivers at Suncor that were automated out of jobs were transferred, but it also eliminated 500 high paying jobs with a low barrier to entry for the next generation and the only benificary is Suncor.
Obviously there are some great things that will come from AI but I see far more bad.
Oh don't get me wrong I wasn't really pointing the finger at you, just a general statement I
Yeah I just don't agree. In the case of art, if any one can create it with zero skill it has no value on top of it being mostly stolen from talented artists. As far as other industries, displacing people doesn't bring any good into the world. They move to industries less affected by AI and drive the value down there. I see a lot of AI as a tool to devalue skills and make money for those at the top.
All the dump trick drivers at Suncor that were automated out of jobs were transferred, but it also eliminated 500 high paying jobs with a low barrier to entry for the next generation and the only benificary is Suncor.
Obviously there are some great things that will come from AI but I see far more bad.
The bolded is a distribution of wealth problem not an automation problem. In general there is a certain amount of human labour required to run society. Reducing that amount allows more leisure. How that leisure is distributed is the role of government. Also in Canada the public pension plans own large swaths of these companies so benefiting Suncor is benefiting people.
Not to mention this should reduce fatalities at Suncor.
The path forward needs to be UBIs slowly increased and taxation of the profits of automation rather than just giving it to the producer and consumer.
The Following 3 Users Say Thank You to GGG For This Useful Post:
The bolded is a distribution of wealth problem not an automation problem. In general there is a certain amount of human labour required to run society. Reducing that amount allows more leisure. How that leisure is distributed is the role of government. Also in Canada the public pension plans own large swaths of these companies so benefiting Suncor is benefiting people.
Not to mention this should reduce fatalities at Suncor.
The path forward needs to be UBIs slowly increased and taxation of the profits of automation rather than just giving it to the producer and consumer.
Suspect a lot of this job loss stuff is wildly overhyped. In our lifetimes - the internet went from basically non-existent to impacting essentially every part of our lives and jobs. Whatever jobs were lost due to the internet were replaced with new jobs created by the internet.
Its possible these AIs will be able to replace jobs, but still a long way away because so much of work is inter dependent on other systems.
I think we'll first see AI taking over call centre and online chat customer service. Customers may actually end up preferring to deal with an AI agent. I know they often have scripted chat bots these days, but they typically suck. I imagine these actually account for a lot of low level, low education required employment. This will also happen very rapidly as options become available, as it will mean massive cost savings. Elimination of jobs for the lower educated that have low barriers to entry is going to have a pretty massive effect on employment numbers.
I suspect this will sound a warning bell for many other industries, and governments will get off their butts and figure out how to deal with this. It makes me think perhaps the best solution early on would be to tax companies heavily for this type of automation, such that they can still save some money, but much of the savings goes to a form of UBI, not to profits. There will be a lot of resistance to this, obviously, but I jut don't see a way around it. And these are not 20 year out problems, this is all less than 5 years away.
I suspect there will be investment to try to make it happen, but I think even the pro chatGPT people here will say there is a fair bit of prompt chasing to get the answer you want. That's fine if you're trying to get it to create code or whatever for you, but if you just want to figure out why your internet is out - customers will get annoyed by it if it isn't giving the answers you want at a very high rate. Quite possible that by having it become an expert on a smaller set of data - it will be able to nail things quicker.
And also very likely that companies will just do it anyways because they use terrible chat bots and phone systems now to try to divert calls from real people. So assuming the cost isn't crazy - they'll do it even if it is annoying to customers to start.
We'll see how quick that happens? I suspect there will be investment to try to make it happen, but I think even the pro chatGPT people here will say there is a fair bit of prompt chasing to get the answer you want. That's fine if you're trying to get it to create code or whatever for you, but if you just want to figure out why your internet is out - customers will get annoyed by it if it isn't giving the answers you want at a very high rate.
Sure, but companies will extend training of the models on their data. And people will typically be calling in(or in a lot of cases, doing it online) for simple information. I think back to many chats I've had with say, Shaw, and an AI could handle the interactions. "Cancel this, how much is that, what's the contract condition for..."
I wouldn't underestimate how quickly companies will move on this when they see cost savings. I wouldn't be surprised if it is 50:1, replacing a call centre employee with an AI.
I also wouldn't underestimate how often people and companies will accept lower quality work that costs almost nothing vs. professional work that costs a lot.
Look at something like Canva. Most small businesses will gladly have a staff member throw something together using that rather than hiring a graphic designer because the marginal benefit of the latter often isn't worth the extra expense.
We'll see the same thing with AI and most types of generated content (writing, graphics, etc.). Even in situations where the finished product is inferior, AI will likely win out because it's cheaper and significantly faster. And it'll become more user friendly once it's integrated into things like Office.
The Following User Says Thank You to opendoor For This Useful Post:
I think one of the worst parts of AI(at least these initial rounds) will be the general "sameness" of everything. You will be reading articles that are all AI generated, with AI images accompanying them, and AI ads through the article. Articles will mostly have the same structures.
I've already found it tough to find good tool reviews and comparisons that aren't just crap text assembled from product specifications. You can sift through pages of Google results, and none of them are written by someone who seems to know what the tool does. It's just content for the sake of content. It's awful. I suspect the easy of generating stuff like that will make it much worse.
The Following User Says Thank You to Fuzz For This Useful Post:
Oh don't get me wrong I wasn't really pointing the finger at you, just a general statement I guess.
Yeah I just don't agree. In the case of art, if any one can create it with zero skill it has no value on top of it being mostly stolen from talented artists. As far as other industries, displacing people doesn't bring any good into the world. They move to industries less affected by AI and drive the value down there. I see a lot of AI as a tool to devalue skills and make money for those at the top.
All the dump trick drivers at Suncor that were automated out of jobs were transferred, but it also eliminated 500 high paying jobs with a low barrier to entry for the next generation and the only benificary is Suncor.
Obviously there are some great things that will come from AI but I see far more bad.
Seems more like a complaint about corporate capitalism rather than a complaint about AI.
No doubt there will be jobs lost to skills displacement via AI, and there will be new jobs created along with new opportunities for self employment. The biggest concern I have with AI as it relates to unemployment is more just about the speed of it all. The risk is in the near term and about the speed of advancement in skills displacement being much greater than the speed of labor force reskilling and having frictional unemployment become permanent. It could be hard for a lot of people to find their way through to what works next.
Of course, those same people losing their jobs will probably actually end up making use of AI to solve their problems of upskilling and being productive again, while corporates on the other hand will always be looking for the next opportunity to increase profits with improved efficiency.
As GGG pointed out, the problem of AI and the workforce is a problem of wealth distribution. I would say it's more particularly a problem of ownership distribution for the channels through which wealth is generated. At the end of the day, AI is just another set of tools to be used or discarded as needs be, and from a certain business perspective employees are just tools to be treated in the same way. That is less a problem of the tool than it is a problem of the social, economic and legal systems.
__________________
"If stupidity got us into this mess, then why can't it get us out?"
The Following 2 Users Say Thank You to JohnnyB For This Useful Post:
Seems more like a complaint about corporate capitalism rather than a complaint about AI.
No doubt there will be jobs lost to skills displacement via AI, and there will be new jobs created along with new opportunities for self employment. The biggest concern I have with AI as it relates to unemployment is more just about the speed of it all. The risk is in the near term and about the speed of advancement in skills displacement being much greater than the speed of labor force reskilling and having frictional unemployment become permanent. It could be hard for a lot of people to find their way through to what works next.
Of course, those same people losing their jobs will probably actually end up making use of AI to solve their problems of upskilling and being productive again, while corporates on the other hand will always be looking for the next opportunity to increase profits with improved efficiency.
As GGG pointed out, the problem of AI and the workforce is a problem of wealth distribution. I would say it's more particularly a problem of ownership distribution for the channels through which wealth is generated. At the end of the day, AI is just another set of tools to be used or discarded as needs be, and from a certain business perspective employees are just tools to be treated in the same way. That is less a problem of the tool than it is a problem of the social, economic and legal systems.
I'd think AI would be more likely than most technological advances to hit higher paying jobs (although not the top jobs in the those markets). But things like paralegals, accountants, analysts, etc will likely be impacted.
I'd think AI would be more likely than most technological advances to hit higher paying jobs (although not the top jobs in the those markets). But things like paralegals, accountants, analysts, etc will likely be impacted.
Yeah, most likely to hit jobs that don't require physical presence and non-repetitive physical interaction with the environment. Things like childcare may be among the most difficult for AI to displace, but could still face pressures as more people may want to compete for those jobs if they lose work elsewhere.
A lot of the people working higher-skill jobs may also be the people with the skills to make their own use of AI to develop other income sources.
__________________
"If stupidity got us into this mess, then why can't it get us out?"
I'd think AI would be more likely than most technological advances to hit higher paying jobs (although not the top jobs in the those markets). But things like paralegals, accountants, analysts, etc will likely be impacted.
On the positive side of things, it will allow accountants the freedom to pursue their true calling in lion-taming.
If more and more of the internet becomes AI generated content, doesn't that mean the AI just starts relying on previous (and worse) AI more and more and we get some kind of AI echo chamber?
I don't think the humans are going away, it's just once again changed the part they'll play, as with every other major innovation that took our jerbs.
If more and more of the internet becomes AI generated content, doesn't that mean the AI just starts relying on previous (and worse) AI more and more and we get some kind of AI echo chamber?
I don't think the humans are going away, it's just once again changed the part they'll play, as with every other major innovation that took our jerbs.
Yes, which is why training AI's is expensive. You need people to rate datasets. So much of the internet will probably make AI's worse if they aren't limited to some sort of source verification, or ranking.
