Traveling into the future

I have a time machine. Unfortunately it works very slowly - The journey takes me 1 second into the future for every 1 second of elapsed time... But I also have two navigation tools which help me comprehend the future as I arrive -- first, I have the past which can often provide lessons about what the future will be like (although not always accurately). Second I have the ability to evaluate what is happening in different places, for different people, but all at the same time ("The future is already here -- it's just not evenly distributed" - William Gibson).

On the first point, I consistently return to reflecting on the "dot com" era sometimes called positively the "dot com boom" and sometimes negatively the "dot com bubble" -- the period roughly between 1995 and 2000. I lived through that period as a creator, entrepreneur, and beneficiary... I also made a few wrong calls like shorting Yahoo! in Q4 of 1998 (I got out before I lost everything...)

There were a couple of important things that I learned during this period which I think are applicable to where we are right now with the AI boom (or AI bubble if you prefer to be glass half full). In no particular order:

• Don't second guess the market - and short positions are terrible when irrational exuberance takes over... you stand to lose a lot of money.
• Irrational exuberance is based on something real, it doesn't just come out of thin air. In December of 1995 the estimated number of Internet users was about 16 million people. By December of 2000 that number had grown to roughly 360 million. We all had graphs showing our investors that the number of Internet users would continue to grow, as would faster connections and more hours online per user (all indicators of the value creation potential of our businesses). If anything we underestimated the growth rate. Irrational exuberance wasn't about getting this growth rate wrong, it was about people investing too far ahead of that growth.
• It is difficult to imagine how a technology will really be used until you are actually using it consistently enough to see the trend lines and even then it is way too easy to lean back on how things used to be before the technology existed and believe that something new won't work... for example -- "People won't buy clothing online, they want to try things on."
• Technologies have a compounding impact and enable new creations that were previously unimaginable -- Internet plus smart phone enables Uber for example.

So where are we on the AI curve? First of all very early (so don't short the market), second of all the projections are probably conservative, third we are just beginning to use this technology enough to imagine how it can be used, and fourth we haven't really begun to see the compounding impact (AI plus humanoid robotics = ??).

This set of things from the past can be subjected to my second tool for comprehending the future (let's just start calling it Gibson's Law). What can be said about these four insights based on looking around right now in the pockets where the future is already here?

• Boy I would hate to be the guy who shorted Nvidia at 1.5 trillion market cap... Is it at the top yet? I learned my lesson, short positions in a boom have unlimited downside...
• There is something real happening on which the exuberance is based (I personally don't think we've reached irrational yet). I haven't tried to draw the graph and I'm not sure what the measure should be -- probably not simply number of users as with Internet adoption. I think it should be some measure of benefits derived, but I don't know of a simple KPI to use (if you do, leave a comment!). I think one leading indicator is in the software industry where the impact of coding tools is having a tremendous impact on productivity especially for junior software developers who can now contribute at several levels above what they had before having these tools.
• We still don't have a lot of data on how the technology will be used but the anecdotal evidence is that companies are finding use cases with productivity impacts way beyond any prior technology -- 99%+ improvements. While we often talk about this as "generative AI" which leads to an assumption that the technology is used to create, the most impactful use cases that I am seeing are analysis, summarization, categorization, and explanatory.
• The early evidence on combinatorial innovation is actually staggering. Combining AI with physical objects (not just humanoid robotics) is going to create entirely new classes of appliances, tools, vehicles, and other devices. A plethora of new services are beginning to emerge with "agentic systems" mediating our consumption and analysis of information, supporting our decision making, and facilitating the actions that are the outcomes of those decisions.

So what can I predict about the future that my time machine is taking me into? First, it is going to feel a lot more strange before it starts to feel normal. Second that with the number of smart people and dollars chasing down the hard problems, we are going to have plenty of solutions and the impact will rapidly grow. Third that we haven't even imagined yet the really big products and societal changes that are coming. And lastly that humanity will likely invest well ahead of the benefits at some point.

But not yet.