Most timely article, Bryan. Ted Gioia recently posted similar thoughts...
"As you can see, a real tech breakthrough grows at a ridiculously rapid pace in its early days. Look at how fast people adopted radio or the smartphone or electricity. And these required huge investments by consumers.
But they’re giving AI away for free at Bing—and it’s not growing at all.
This is not how consumers respond to transformative technology. The current demand pattern resembles, instead, what we would call a fad or craze.
Unusual piece Bryan but well worth the read. One major factual error. OpenAI is a Not for Profit and is not owned by Microsoft. The have a deal with Microsoft which is non0exclusive and use them for infrastructure through Azure. ChatGPT losing users to 1.5 billion is explained by not only the API but other products - Claude, Bard etc. Also, Facebook, and other large tech companies do release Opensource AI e.g. facebook and LLama. The idea that this will fade into insignificance seems bizarre to me for all hte resons outlined in the Shindig session... but good piece.
Thank you for the careful look, Donald. Yet doesn't MS own a majority stake in OpenAI?
I'm not sure of the probability of such an event But it's an open possibility now, and I'm not seeing folks talk about it. As a futurist I have to keep possibilities in play.
Now, the open source alternative - that's one I want to dive into more. What's the best kindergarten starting tool, Red Pajamas?
Deal not entirely clear but there is no equity stake by Microsoft in OpenAI. It's is thought to be a 'recoup through profits' deal with a minority equity stake later after repayment. It retains its Not-for-profit status.
Deal is likely to much more complex as Microsoft are also using OpenAI tools and models. There has to be a reckoning. Fact that Sam Altman has no equity is interesting and quite unique. OpenAI needed moonshot funding, to build and train the model, as well as ability to distribute for free. For that they needed a Microsoft, Google, Amazon or Chinese entity. The last is out for obvious reasons, while Google have different objectives as do Amazon. Deal made and makes sense. The expectation among many that brilliant IT services should be free seems unrealistic, so I'm happy to pay $20 per month. The Fremium model has emerged as one that is fair, as majority can use free service and extra functionality, namely Plug-ins, better model etc has to be paid for. Compared to costs of HE, that's an absolute steal.
Powerful and interesting article. Thank you for your work on this Bryan. I think it is very important that we think things through and consider all possibilities so as to be ready. Yet in my analysis of all of these aspects I do not believe it is even possible for AI to go away in any major way. Both its adoption and utilization has been record setting. It saves times which translates to saving money so businesses will not let it go. In the same way it greatly enhances people's (students, instructors, all people's) ability so they will not let it go either. I fully believe the decline seen for ChatGPT has been due to the summer break from school and the increase use of API's within organizations.
I totally do agree though that environmental concerns and regulations need to be addressed, but there simply is no putting the genie of AI back into the bottle. AI is part of the 4th industrial revolution and will continue to propel us into the future.
Assumption here is that the US courts have global reach. It's a big world out here. Not all models are US. If you are going to do this, see AI as a global issue. The Chinese are on a par. In truth, there is ni way the US will consign themselves to economic decline as others make the productivity gains on offer.
Regarding the economics of the industry, this report is eye-opening...
www.washingtonpost.com/world/2023/08/28/scale-ai-remotasks-philippines-artificial-intelligence/
To what extent are LLMs versions of the old Mechanical Turk?
Most timely article, Bryan. Ted Gioia recently posted similar thoughts...
"As you can see, a real tech breakthrough grows at a ridiculously rapid pace in its early days. Look at how fast people adopted radio or the smartphone or electricity. And these required huge investments by consumers.
But they’re giving AI away for free at Bing—and it’s not growing at all.
This is not how consumers respond to transformative technology. The current demand pattern resembles, instead, what we would call a fad or craze.
And this is just one warning sign among many."
https://www.honest-broker.com/p/ugly-numbers-from-microsoft-and-chatgpt
Thanks for your post! And I agree with the points that Brent and Don made. Here’s an interesting take that makes sense to me. https://open.substack.com/pub/synthedia/p/the-imminent-death-of-chatgpt-and
Unusual piece Bryan but well worth the read. One major factual error. OpenAI is a Not for Profit and is not owned by Microsoft. The have a deal with Microsoft which is non0exclusive and use them for infrastructure through Azure. ChatGPT losing users to 1.5 billion is explained by not only the API but other products - Claude, Bard etc. Also, Facebook, and other large tech companies do release Opensource AI e.g. facebook and LLama. The idea that this will fade into insignificance seems bizarre to me for all hte resons outlined in the Shindig session... but good piece.
Thank you for the careful look, Donald. Yet doesn't MS own a majority stake in OpenAI?
I'm not sure of the probability of such an event But it's an open possibility now, and I'm not seeing folks talk about it. As a futurist I have to keep possibilities in play.
Now, the open source alternative - that's one I want to dive into more. What's the best kindergarten starting tool, Red Pajamas?
Deal not entirely clear but there is no equity stake by Microsoft in OpenAI. It's is thought to be a 'recoup through profits' deal with a minority equity stake later after repayment. It retains its Not-for-profit status.
So OpenAI would have to pay MS back from whatever extra money it earns?
Deal is likely to much more complex as Microsoft are also using OpenAI tools and models. There has to be a reckoning. Fact that Sam Altman has no equity is interesting and quite unique. OpenAI needed moonshot funding, to build and train the model, as well as ability to distribute for free. For that they needed a Microsoft, Google, Amazon or Chinese entity. The last is out for obvious reasons, while Google have different objectives as do Amazon. Deal made and makes sense. The expectation among many that brilliant IT services should be free seems unrealistic, so I'm happy to pay $20 per month. The Fremium model has emerged as one that is fair, as majority can use free service and extra functionality, namely Plug-ins, better model etc has to be paid for. Compared to costs of HE, that's an absolute steal.
I agree with most of this, from the steal to your sense of the moonshot.
Now that the stuff is in open source, I wonder if there's a chance to break out.
Should Higher Education 'break out' into 'open source'?
Powerful and interesting article. Thank you for your work on this Bryan. I think it is very important that we think things through and consider all possibilities so as to be ready. Yet in my analysis of all of these aspects I do not believe it is even possible for AI to go away in any major way. Both its adoption and utilization has been record setting. It saves times which translates to saving money so businesses will not let it go. In the same way it greatly enhances people's (students, instructors, all people's) ability so they will not let it go either. I fully believe the decline seen for ChatGPT has been due to the summer break from school and the increase use of API's within organizations.
I totally do agree though that environmental concerns and regulations need to be addressed, but there simply is no putting the genie of AI back into the bottle. AI is part of the 4th industrial revolution and will continue to propel us into the future.
Thank you for the thoughts, Brent.
So let's take this a little further. If a court shuts down the big firms, how might consumer demand be met?
Assumption here is that the US courts have global reach. It's a big world out here. Not all models are US. If you are going to do this, see AI as a global issue. The Chinese are on a par. In truth, there is ni way the US will consign themselves to economic decline as others make the productivity gains on offer.