I think the hype cycle (the name for this chart) is inevitable because the hype cycle is a human phenomenon, not a tech one.
Yes it applies to the internet and to Web3, but it is also a good emotion map for anything exciting in life.
If I were to give that chart a title of “new girlfriend/boyfriend” or “new job” or “your team signs a new star player” it would still work quite well.
Whenever we humans encounter something novel we always get over excited about it, form unrealistic expectations and then form resentment when the thing can’t reach those un realistic expectations.
Over time, we start to appreciate the many good qualities and forget about the unrealistic expectations we had at the beginning.
I think this is an incitement point Chris, we often get a huge spike of dopamine when something // someone new enters our lives and then, dopamine crashes back below baseline before reverting to our mean state.
It’s definitely applicable to hype cycles in new technologies as well.
Just like the internet in its infancy, AI will have to find a way to make money. The 99-2000 bubble was the initial wake up call for the early days of the internet. This was the “show me the money” event of its time. Like early Cable TV, somehow these new technologies needed to be monetized. Cable quickly realized that just charging a monthly fee wasn’t enough, so advertising was added as well. The internet followed a similar path where now your online newspaper requires a subscription, but you have ads as well We will all have to see what AI does for its monetization model.
Completely agree. Since there is a direct conflict of interest between the sourcing information and advertisers which will pay to be listed higher (Google), a subscription model is preferable.
Which means people will have to find enough value to justify $20 a month, and also, like streaming, one or two of these companies will run away with the market, and the others will have 40-50% monthly churn, unless they can create moats
AI is not the internet, though. I think what made the internet so big is that it was a communications technology, like phone lines but better. I’m not sure if the comparison holds true for AI.
Haha, that's all but fair. Although I feel the onus on those making the positive claim to provide evidence for its transformational utility?
Also, there are some great articles being written recently about the AI CapEx infrastructure build out and the game theory behind it (https://www.sequoiacap.com/article/ai-optimism-vs-ai-arms-race/). As a bystander in all of this, I'm looking at this and thinking: here are a handful of companies investing in huge data centers to train similar models, of similar sizes, with similar capabilities. What are we really doing here? Is this really the way to go about it?
On that last point we’re completely aligned. LLMs seem like the most inelegant, brute force way to solve this problem, and I hope that there are stealth companies out there at least trying different architectures
Great take! I prefer to think we're in the trough of disillusionment so I'm not sad.
That’s a great way to put it! Trough of disillusionment 🤝
I think the hype cycle (the name for this chart) is inevitable because the hype cycle is a human phenomenon, not a tech one.
Yes it applies to the internet and to Web3, but it is also a good emotion map for anything exciting in life.
If I were to give that chart a title of “new girlfriend/boyfriend” or “new job” or “your team signs a new star player” it would still work quite well.
Whenever we humans encounter something novel we always get over excited about it, form unrealistic expectations and then form resentment when the thing can’t reach those un realistic expectations.
Over time, we start to appreciate the many good qualities and forget about the unrealistic expectations we had at the beginning.
I think this is an incitement point Chris, we often get a huge spike of dopamine when something // someone new enters our lives and then, dopamine crashes back below baseline before reverting to our mean state.
It’s definitely applicable to hype cycles in new technologies as well.
Thanks for reading!
I am not sure we have reached the “trough of sorrow” yet, but we are getting there.
Still, I find these technologies disruptive. When was the last time you Google’d anything?
LLMs make better search engines, almost all of the time.
I agree, we’re approaching it now // shortly. But completely agree, the tech is here to stay. Perplexity alone is 10x better than Google search.
Just like the internet in its infancy, AI will have to find a way to make money. The 99-2000 bubble was the initial wake up call for the early days of the internet. This was the “show me the money” event of its time. Like early Cable TV, somehow these new technologies needed to be monetized. Cable quickly realized that just charging a monthly fee wasn’t enough, so advertising was added as well. The internet followed a similar path where now your online newspaper requires a subscription, but you have ads as well We will all have to see what AI does for its monetization model.
Completely agree. Since there is a direct conflict of interest between the sourcing information and advertisers which will pay to be listed higher (Google), a subscription model is preferable.
Which means people will have to find enough value to justify $20 a month, and also, like streaming, one or two of these companies will run away with the market, and the others will have 40-50% monthly churn, unless they can create moats
AI is not the internet, though. I think what made the internet so big is that it was a communications technology, like phone lines but better. I’m not sure if the comparison holds true for AI.
Good point, let me put it like this. Massive infrastructure shifts usually swallow the previous medium:
Television swallowed radio and incorporated it into its infrastructure layer
The internet swallowed movies, tv, and radio and incorporated them into its infrastructure layer
AI has swallowed the entire open internet (including video, text and audio) and has incorporated it into its infrastructure layer.
I don’t have statistics, but I’d assume less than 25% of people today use the internet strictly for communication.
The main use of the internet today is to query information and ingest text and video
AI is the natural successor of the internet in this way
Is it though? I know people what it to be that. I’m more skeptical.
You haven’t provided any evidence or rationale to the contrary, so forgive me if I take your skepticism with a grain of salt.
Haha, that's all but fair. Although I feel the onus on those making the positive claim to provide evidence for its transformational utility?
Also, there are some great articles being written recently about the AI CapEx infrastructure build out and the game theory behind it (https://www.sequoiacap.com/article/ai-optimism-vs-ai-arms-race/). As a bystander in all of this, I'm looking at this and thinking: here are a handful of companies investing in huge data centers to train similar models, of similar sizes, with similar capabilities. What are we really doing here? Is this really the way to go about it?
On that last point we’re completely aligned. LLMs seem like the most inelegant, brute force way to solve this problem, and I hope that there are stealth companies out there at least trying different architectures