Thanks Alex for sharing your take on what could be the "potential" roadmap for OpenAI and ChatGPT similar to AWS and iPhone success stories, you have also mentioned how it could fall behind and impact its partners as well.
My question is: can these companies survive the gigantic costs in compute cycles to keep these AIs running in a performant fashion, once they become endemic to all these tools and operating systems. How do you even begin to recoup that cost, when people are becoming much more resistant to the "dozens of small subscriptions eating my wallet" model? Even with improvements in efficiency, it costs millions of dollars per training cycle for an LLM; how will they cover adding tens of thousands of GPUs a month, to get this in front of many users as they claim to want?
What if an AI chatbot fatigue occurs in the coming months? I wonder how Big Tech firms would adapt.
Thanks Alex for sharing your take on what could be the "potential" roadmap for OpenAI and ChatGPT similar to AWS and iPhone success stories, you have also mentioned how it could fall behind and impact its partners as well.
My question is: can these companies survive the gigantic costs in compute cycles to keep these AIs running in a performant fashion, once they become endemic to all these tools and operating systems. How do you even begin to recoup that cost, when people are becoming much more resistant to the "dozens of small subscriptions eating my wallet" model? Even with improvements in efficiency, it costs millions of dollars per training cycle for an LLM; how will they cover adding tens of thousands of GPUs a month, to get this in front of many users as they claim to want?
AI could swallow up the companies if they've spent too much resource and time on it.