Impact of AI

It seems clear we are headed towards a mammoth amount of spend on AI infrastructure as the big players are feeling sufficiently comfortable with training workflows and consumer uptake, probably quite enticed by the ginormous market opportunity as between PwC and OpenAI.

This will put a lot of pressure on land, powergrids, and others, but might not be all bad:

• Building up compute should be flexible. The exact workloads are envisioned to change, but that seems easier to do with a datacenter than with other pieces of infrastructure
• Renewable energy is set to be a part of this, including wind and solar.

Ok, but what’s the purpose?

Assuming the datacentres are up and running, preferably on green energy, Is AI-driven automation going to devalue work to 0?

Sure, for some text-based applications where you can afford to not have it rigorously controlled (and in such a case you might question what the value is, if it can run without human oversight, we are back to the 2 LLMs having a great long conversation by themselves, with limited impact)

In the public market lens with its implied valuation, there are companies losing out, others doing okay and one big winner.

1. Salesforce, UIPath etc. Are they already displaced by AI? Maybe, or maybe their own efforts are not yielding anything. General perceptions matter.
2. The big companies as before: quartely, all seems fine, but layoffs keep happening
3. NVIDIA is obviously the big winner. The H100 chips that are best able to do the work of creating AIs are seen as the be all end all, and jealously guarded, but these too have an expiry date when the next chip comes along.

Should we dream bigger?

As for examples of AI-driven disruption?

• Bluntly, full self driving is a pipe dream. Everyone’s posted 1 picture of a vacant steering wheel in a car inching through some SF neighbourhood, but there’s still millions of drivers everywhere doing all kinds of crazy stuff, and AI is not going to replace that anytime soon, let alone join the human majority on the road as an autonomously operating vehicle.

• GenAi creates weird dynamics on platforms as content gets skimmed off human writers looking to break through, and then mulched back into a reformulated version, concievably in a perpetual loop. I suppose now these types of experiments can be run without an activated base by virtue of users already on the platform and the self-starting aspect of GenAI, but the experience is lacking as there isn’t any trust yet on what this thingy is actually doing. ChatGPT may be the exception rather than the norm in product, and it has now plateaued at 100M users per month after a breakneck-speed entrance.

So, what next?

Now, can AI help to grow us out of the current crisis and into an abundant future? Answer: not if the goals remain the same as they are now!

We need to use this tech to do something truly different and somehow irrational from our current point of view.

What does that look like? I don’t know. But!

• I am pretty sure it doesn’t involve an AI usage that can be easily copied, intercepted, bureaucratised. The example here is Ai for pitch decks, which really baffle me, and seem to be focused on precisely the wrong approach of fitting to a previous pattern.
• Instead, we humans need to:
  • be clear-eyed about the tradeoffs made today, but the current crisis needs to compel to action
  • take a future-proof approach: not what works now but what can truly last
  • embrace a mindset of diversity to spot when we’re going off course

And at susty.vc, our goal is to find and support those going for this approach.