The last 9 months we've seen a boom without historical precedence. I am of course talking about the ChatGPT and AI boom. The boom is over now, and ...
For further actions, you may consider blocking this person and/or reporting abuse
Not sure if I agree with you here. Your claim is that the ChatGPT Boom is over. What data do you use to support your claim? Are there any studies on ai startups founded vs ai startups closing? What about the rate at which the technology is adopted?
I get it that the dust settles a little, but we are far from getting rid of those schemes you mentioned or even ai plowing it's way through our work environments or society.
There is a concept of the Gardner hype cycle that explains that on every hype there is a phase of adopting or dying. And I think we will enter this phase rather soon. But there are yet some milestones to come, that eventually will restart a hype (i.e. microsoft releasing their business copilot that affects 100.000 knowledge workers world wide).
Might only be me, but I read some frustration between the lines. I wished you used more evidence to support your claims and be more neutral, when approaching this ticket as this is an important ticket we all should discuss. Sadly there is no room for actually discussing the subject here :/
We're on the downwards slope of the Gartner (not Gardner) hype cycle according to measures of daily users of ChatGPT. This has been documented to DEATH by multiple independent sources, including OpenAI themselves.
If you believe in Gartner, and the existing research, it'll drop down for some 2 to 3 months now, at which point all the "get rich fast scams" will automatically weed themselves out, while the technology matures, and starts going (slowly) back up again, producing a more healthy environment.
This is a GOOD thing for anybody serious about AI and Machine Learning ...
OK, care to elaborate ...?
If you're just looking to disagree because you want to "win the argument" instead of having an actual discussion, with real arguments, I will block you yes, faster than you can spell B-L-O-C-K. And if you ask for "a source", I will also block you. You can Google like the rest of us I presume.
If you however want to have an intelligent discussion, without ad hominem arguments, and false accusations (you're close to this already, consider this a warning), I will also block you.
I'm simply too old to argue with people looking to "win the debate" ...
If you're looking for a punching bag, go somewhere else 😊
Then I apologize and you are of course more than welcome to post your opinions, stats, and links. However, when it comes to references? 500 Facebook groups citing Bloomberg, Gartner, New York Times, "whatever", in addition to equally many groups at LinkedIn, in addition to OpenAI, Sam Altman, and every single VC fund on the planet.
It would be impossible to miss 😉
My hostility related to "show me references" originates from (the fact that) it is a well known control techniques. This might not have been your purpose, but I'm tired of having to "show references" to things that (should be) common knowledge 😊
I see. Glad we sorted things out :)
Something about maggots crawling out of wood, telling him that. I'll ask ChatGPT about it.
This won't look good, once our AI overlord took over the world :D
Hehe, call me an optimist, but I suspect that also the Terminator cares about quality. Besides, after my own experiences with the actual value of LLMs lately, I'd sell out John Connor for a packet of Jucy Fruit :D
Hahaha :D
I recently posted about amalgam my idea to solve the problem of software development requirements being misunderstood by developers, I’d be curious to find out your thoughts as you have wrote about AI with such gusto for 9 months.
Is my tool worth further development or just hot air in your opinion 😊
I would love to look at it? Link?
However, if you ask questions like this, I suspect it's got value, since your uncertainty resulting in asking questions, becomes what drives you to build better product 😊
If you on the other hand had started out with: "Yes I agree, but OUR stuff it different. With our stuff you can get rich fast without putting down any work" - I'd answer: "Buy NFTs NOW! VC Monkey are BOOMING!" 😂
Thank you I can see your logic that makes sense, I am actually trying to solve a real problem but I hope I haven’t missed the boat
🚤
dev.to/adam_cyclones/never-write-a...
It looks legit. As a general rule of thumb I can tell you to "scratch your own itch". You're probably not the only one in the world with that itch. Then make sure you have fun while scratching it. The last point is crucial, since it results in you spending a lot of time on it, making it better, until it's simply so good they cannot ignore it anymore ... 😉
Particularly the last point resonates, I often wonder where I am going with a project and just stop
If you stop voluntarily, it's not your scratch ;)
Depending on how the ever growing barage of copyright infringement lawsuits against OpenAI plays out, everything built atop ChatGPT may come crashing down like a Jenga tower if the courts decide in favor of the many plaintiffs. There is a real possibility that OpenAI may eventually be forced to scrap their models.
There is also the separate issue, also copyright related, that the US copyright office has indicated and courts are already starting to agree with that the creations of generative AI are not copyrightable because they aren't created by a human. A publisher wants to own the copyright to content. The makers of movies want to own the copyright of scripts. The music industry wants to own the copyright to lyrics and music. None of these industries can make money if anyone can legally copy their books, movies, music, etc. So if courts continue to find that the creations of generative AI are not copyrightable, it will cause many (at least in some industries) to take pause when considering whether or not to utilize it.
It's simpler than you think. If I take something that's in the public domain, and add one character to it, the derived work in its entirety is mine according to copyright laws. You can copy Shakespeare (public domain), and change Hamlet to Snoop Dog, and the copyright is yours ;)
That is not how the copyright of a derivative work works. See "311.2 The Originality Requirement for Derivative Works" of Chapter 300 of the Compendium of US Copyright Office Practices. The derivative work "must possess more than a
modicum of creativity." Another quote from this section "Miniscule variations do not satisfy this requirement." Concerning public domain material contained in the derivative work, it indicates:
You can legally take Shakespeare (public domain) and change name of Hamlet to something else ("Snoop Dogg" would cause issues with his trademark, so let's assume some other name, how about Zyzzyva, last word in dictionary). However, you won't own the copyright to that derivative work if that is all that you do. None of what was copied from Shakespeare can be claimed, which is nearly everything in this case. And you can't even claim copyright on your new character name because character names are not copyrightable (313.4C), although you might be able to trademark that character name.
You're of course correct, but the point was to illustrate that derived works creates new copyright. Typically, for ChatGPT generated stuff, you'll edit some "significant portion" of the original works - Maybe some 10 to 20 percent, at which point the derived work becomes a new unique work.
With code, I know Trolltech used to put the threshold at 5 lines of code in a patch before they needed an explicit permission to use the work.
But yes, changing Hamlet to "whatever" obviously won't work. Changing it to Snoop Dog also (probably) creates trademark issues. But generating AI content, and ensuring you're able to copyright it, by applying changes to it - Is still relatively easy ...
I see your point now. (Step 1) Generate something with A.I., which is not copyrightable. (Step 2) Human transformation of result to whatever extent is necessary to create a derivative that is copyrightable.
That handles that issue, provided step 1 isn't infringing copyright with its dataset. I suspect it will be a while before courts fully resolve that for systems like ChatGPT and some other systems currently being challenged.
But training on data that you own or can license would be a different story. A few years back there was a documentary on Disney+ about how they created a deaged Luke Skywalker for finale of Mandalorian season 2. They used deep learning, but they went through tons of 1980s footage to get the images they needed. Many of which are probably under copyright owned by Lucasfilm, or Disney, or could be licensed from relevant news companies. They didn't discuss that so not clear if they actually licensed any training images they didn't already own. So dataset (probably) free of copyright issues.
They also though went beyond just generate with AI, then modify. Humans were integral throughout process: hand picking training images, two actors on set, humans involved in mapping to actor faces, human camera operators on set, etc. So although A.I. was used in deaging, there was a very significant human element to the creation. Lucasfilm/Disney clearly own copyright to the deaged Luke Skywalker in this case because of that human element.
This step has already been resolved. No copying, no copyright. If they're to ban this, we'll have to ban every single IP created from the day we crawled down from the trees and started painting caves.
They're just not willing to admit it's been resolved yet ... 😉
I suspect you've got a "holy cow" in this debate? Care to tell us what you're working with? My reasons for asking is because I think you've got this wrong. AI won't replace humans, humans with AI will replace humans without AI though ... 😉
If you're working with what I suspect, the above might comfort you - At least that was my intentions for writing it ...
No "holy cow" in debate. I've been in A.I. research for 25 years, and will be until I either retire or die, and I've taught A.I. for 20 of those years.
I'm also interested in IP law. The copyright issue related to OpenAI's training data is definitely not resolved. It includes copyrighted content (copying very much did occur). It is the matter of whether or not their use of that content is infringing. If the NYT actually sues, as they're considering, the potential damage to OpenAI if that suit is successful is far worse for them than the couple suits filed so far by some book authors. Real potential for some high monetary damages, and real potential to be forced to scrap their current models.
I'm willing to bet $100 on that they'll come nowhere! You want to bet against me? 😉
It is too close to a 50/50 outcome in my opinion for me to bet in either direction. I'd never bet on a coin flip.
It's only a flip coin if you don't consider what's at stake. The potential gains for the establishment are simply too large to be ignored. It's a multi trillion dollar invention. Believing the court will stop this, is madness. It's the type of stuff people would go to war for if required ...
But we'll see. I don't have any particular inside information. I just know that where there's money to be made, solutions will be found ...
Perhaps the initial noise is over, but the actual work just beginning. I'm running a small startup in AI and we have been recently working on the applications of GPT (LLMs in general) in different domains.
I certainly don't think GPT is a "thing of the past"; we are just at the dawn of the generative AIs.
Feel free to check out our open sourced code. It's a work in progress, but I assure you it is NOT a 5 min work. lol
github.com/RoboCoachTechnologies/G...
You didn't read more than the headline before commenting I suspect. It's OK, I do too sometimes ...
I read the entire article, and I also posted your article on reddit.
I understand the point that you are making, and I am not disagreeing with the main point of your article. I just made a comment, which I expect to be mostly is inline with your point, and only contradict the title. Just a comment, no hard feeling.
Thank you, and I apologise. My reasons for saying what I said, was because you're echoing my exact words from the article. Yes I agree, the real work is just starting 😉
Absolutely.
Interesting idea :)
Thank you. That differentiates us from the likes of GPT-engineer and Aider.
It's a work in progress though, I must admit. We will make a new release in a few days with some improvements.
Here's what we're doing, that separates us :)
Of course, contrary to most others, we've got ...
We've got some 100+ additional features, but already with the above 4, we've thoroughly smoked 99.99% of every single "competitor" out there 😉
The above one wasn't written by me, but our CEO ...
It's really interesting. Because recently I ran into some innocent people like that and I had to show them the full list of the "official" GPT chat for Hebrew speakers. A partial list:
(Hope I'm not "helping" here for a few more innocent people )
We've got RTL and Hebrew support in our chatbots - Just sayin' ... ;)
The truth is that just now I took a look at this repository of yours on GitHub
We've got "a bajillion" ChatGPT related features in that repository,such as for instance:
We're about 10,000 light years ahead of 99.999% of every single ChatGPT chatbot out there ... ;)
Hey loved this post. And yes, it's a bit cringe while reading all the get rich schemes with ChatGPT, AutoGPT, ChatGPT Plugins and all the NPC tweets that goes like, "ChatGPT is the thing of the past, here are 5 ChatGPT plugins to optimize/enhance xyz and get an unfair advantage."
Countless YouTube videos explaining why AI will take your job, and how to future proof your job from AI. Something like: "Skills that I'm learning that will make my job AI proof", and then they mention coding and web development in it. (Plus, they forget that the same youtuber has made a video on how AI will take web development jobs, they need memory more than these GPT/LLM agents).
Anyways, we have seen the same stuff happening with Web 3 as in 2021. Many invest in crypto get rich stuff was going on.
I think this keeps on going on in cycles, we can use this AI to solve a lot of critical problems, but that's a blog post on its own. But thanks Thomas for this post!
Thank you Saurabh, and yes it is a cycle. It's more of a problem with the human mind than technology though, so I'm not sure we'll fix it unless we crawl backup into the trees again ... 😜
Actually it's not like the ChatGPT boom is over.
Not it's more like using ChatGPT to fool boom is over.
Of course you're right, but that is the definition of "the boom being over" ...
No, I believe the boom refers to all kinds of usage, not just fooling apps.
You are right in saying that the major market for ChatGPT Apps are actually people making basic stuffs and selling them for a relatively high price. But still there are many open source tools/integrations that are coming up everyday. These are also a part of "the boom" which I think isn't over yet.
This is a part of a healthy eco-system where people are trying to legitimately build actual value on top of ChatGPT. With "boom" the meaning is the inflated belief in that ChatGPT will somehow make you rich without having to put down effort. Think "gold rush", where everybody believes they'll become rich without having to put down work.
People trying to build actual value on top of ChatGPT is not a "boom", regardless of how many such you find. This is just people opportunistically putting down actual work on top of the platform.
"Boom" is a financial definition, and it's got a lot in common with "a bubble" and "a gold rush" ...
Thank you so much @polterguy for clearing this out.
I was considering "boom" to be any high usage of something. I wasn't aware about the financial definitions.
Considering this, you are absolutely right that this "boom" is over. And it makes me feel relieved.
Chai GPT?, damn, why haven't I thought of it before?
Hahahahaha 😂
Forget about the X thing, let's do TEA man!! 😂
Lets do it, and then say: "all started with a meme"
This is the best post I've ever read in my life
Thx mate :)
Interesting. I never thought of it this way tbh.
Thank you :)
Great post. Yeah we have been waiting for this.
Thx mate, the psychosis was real yes 😉
Where's your proof?
Seriously? 😂
OK, here 😂