Ask HN: How do you plan on making money with Open AI API?
It seems that compared to DaVinci, the other 3 models are pretty difficult to effectively use. I spent hours trying to get usable results with Curie, but to no avail.
I'm just curious to know from those who are building applications using Open AI APIs, how are you planning on monetizing your applications when it costs 1-2 cents to get a decent result. That is quite expensive.
139 comments
[ 3.3 ms ] story [ 196 ms ] threadYou must be talking about ChatGPT Plus? The GPT-3 API is pretty cheap in my experience
They're already using gpt3 afaik
After that, it's half the price as davinci on paper, but you need only a fraction of the tokens. They've said they plan on making it cheaper in the future too.
The sample is in /ft/ folder, instructions at bottom of Readme. Feel free to drop an issue if it's hard to understand. I copy-pasted the instructions in 3 min so it's missing a few steps.
With that it would be less effort to create fun and useful little apps without needing to charge the user directly.
How likely is it a person will know what open ai is, be willing to set up billing and then also pay you separately?
The UI will set it apart, and more features are going to be built after launch to further differentiate InventAI from other products.
Create a syllabus for learning [insert subject] https://www.youtube.com/watch?v=tVxy1QDYapQ
Granted many companies start off as the first and transition to the second type of company but OpenAI is so far away from that you'll probably have a good 10-20 year run before that.
And when that happens a version 5 or 6 will probably pretty much wipe 80%+ of the white collar workforce including your job. If V4 is just a medium leap, a V5 or 6 will easily take your job.
It's time to be a cook, a carpenter or a nurse.
Personally i have multiple startup projects i'm worried about AI will be able to do easily in 2-5 years, and i can't really think of any projects anymore where that isn't the case?
There's almost something bizarre about the bell curve where IT has been so highly paid and progressed so much that developments are now making the sector potentially eat itself in the long run.
Is this crazy thinking? I hope it is!
While it's already surprisingly good and useful, it still is often not good enough. And we could hit a celling faster than we hope.
And more interesting: just because a lot of people know about it and are able to use it, there are so so much more people who actually don't.
Look on how a lot of companies still struggle with it or digitalization.
Do what? Setup a corporations, handle logistics, pay the accountant, open a bank account? I mean, there's more to running a company than writing the code and fixing bugs (assuming AI will get better and not worst at it as time goes by).
Outside the HN bubble, from an entrepreneurial perspective a programmer is an expensive machine that implements ideas into code. No one cares if it's in rust or cobol, if it runs in O(1) or O(log n), etc.
So in that sense, make some jobs are under threat but most likely something else will come along.
<HERE BE DRAGONS>That's human history and tbh... can't wait for an UBI-based + social welfare future.</HERE BE DRAGONS>
There are high chances that such services become "AI friendly" overtime, as in are easily understood and automable by AI. Kind of like roads, parking, and fuel stations for cars.
if you're a programmer working now in most places, and in the future you are going on UBI, you will probably be going bankrupt because UBI will not be paying anywhere near what you are earning now.
Not working because anything you can do automation can do better for cheaper, will get you basic income and basic decent living conditions.
yes which for most of the people on this site would probably be a big decrease in living standards and also have possible debts etc. that cannot be paid off by UBI because people thought I am programmer, programmers get paid good. So in short if you lose your job because automation does it better, even if you get UBI you are probably screwed.
And if the living standards of a minor fraction of very well off people decrease and the living standard of many people increase, is that so bad?
Hypothetically, when someone is going from from say 100k to 90k salary per year and another from 0 to 10k of UBI, what has a bigger impact on someone's well-being do you estimate?
Of course it'll never be this clear cut, but for the sake of argument.
If the State provides financial assistance (Housing! Medical Care!) I think most people could survive.
I have some spicy spicy takes about that future. If you peruse my other comments on my profile you’ll see them.
I’d expect the most likely outcome to be competiting welfare systems. The largest will be public and be extensions of the existing welfare systems. Private entities— corporations, nonprofits and billionaires- will run combined housing and UBI projects. You get to live with other people who have declared loyalty to the same entity. It will be like serfdom in the middle ages, but without the work.
Depending on who you live under, it could either be really good or really bad. In the suckier places it can be like a prison or a cult.
Here’s some food for thought. The State already runs living spaces for people who usually aren’t economically productive. Those spaces are called prisons and housing projects.
If communist societies could give everyone an apartment we can do the same with all the productivity unleashed by AGI. The tougher question is how we do that without abusing the tenants.
Maybe we can tame the dragons.
Writing prompts and reading the wrong or incomplete code is harder than typing code. The only thing it saves you is this initial moment where you literally have no idea how to get started.
Also, cooking and carpentry have been automated already.
I don't know why this pattern keeps repeating. ChatGPT is a language model which means it is good at manipulating language. It generates text and since programs are text, therefore it makes "programmers" redundant because programmers just generate text. That is the kind of logic that you are using.
As I said above, if you just treat AI as a human language to code compiler then your source code is the words you put into the AI and now you need to pay the cost of getting the prompt that solves your particular problem.
So now you have cost of prompt writing + X% failure chance multiplied by the cost of writing code yourself Vs cost of writing code yourself and again I am just talking about writing text. I do more things than just write text, like almost any software developer out there
It's like those movie summary channels idk why those get under my skin... just a waste.
Oh here's an example Chrome Lords 1988 on YT saw this one recently
Maybe I'm annoyed because it works. You take a bunch of public things like public images of airplanes and make several channels/videos on each plane and facts... you could automate this and eventually something will work. The TTS voice from medium in particular (guy voice).
I really liked this video, I see your point!
2. (not me personally) Tuning and personalisation of LLMs. - As a company, I want private data to be combined with internet-scale data so that I can get actionable data - As a company, I want internal identity management to secure who is told what by the LLM that is trained on private data.
3. (not me personally) Quick buck. Skin + API call + marketing = short term cash until people clock on that it is cheaper elsewhere.
It reminds me of the browser. With browsers there is no money making move but you can use it to improve your work. That and you can't trust the output. Someone needs to babysit it.
But we are evaluating GPT3 for earnings call summaries and comparisons.
https://kadoa.com allows you to extract any data from any website in the format you want.
(I say this because that's the way to make money from Google search).
All jobs that use natural language as a tool or resource are fair game.
Well I think you've all raised some really interesting points, maybe we should circle back and think about the new issues listed in the risk register
Ahem ... I am a large language model trained by OpenAI and I don't have the ability to ...
I don’t use it myself so can’t really give a good review but my window into the world of long term users would suggest that saying you were cut off wouldn’t strike anyone else as unusual.
I just came up with a great title for a management book “The Alpha Zero Manager”. Now I’ll just ask GPT to write the content.
[1] https://twitter.com/explain_paper/status/1620185211127595008...
AI aided shopping "experience"
https://www.kadoa.com/
Probably because that's a lightning-strike sort of event that you can't really plan on.