Robotics - Bigger, Better, Faster, Sooner and Quite Different than what you think
It's Accelerating
There are a dozen amazing things happening in robotics, including the below item from NVIDIA and orders of magnitude improvement in evolutionary design acceleration coming out of Northwestern Uni. The deep and broad leverage of AI advances into this field is just out of the starting gates in part because there will be more advances in AI in the next three years than in the last fifteen. Very clever stuff is going on with Soft Robots, Cobots, mobility, teaming, common sense for robots (see Cosmos-Reason1), plus more esoteric things like ARD (Amphibious Robot Dog) and twisted liquid crystal elastomer-based soft robots to transport things through the air. Yep, that's a thing from very smart folks at North Carolina State University .
The Distant Future - 2028 - Big Broad Impacts
In three years, we will look back and see a shift that will look like we went from dreaming to doing ,flashy to pragmatic, from isolated/noteworthy to broadly distributed and commonplace.
There will be small, smart robotic solutions that simply improve outcomes dramatically while reducing cost structure, often through variable cost impacts and quality improvements. These solutions can be thought of as defect and rework elimination plus cycle time gains.
With the tailwinds of AI, this will exceed all expectations. Robots will also be near real-time self-assembling for tasks as they arise or are scheduled. They will have functional LEGO for tasks, mobility, vision intelligence, scale, and speed. The start of the day and peak hour in a warehouse operation are very different, for example. Robots will live inside an Agentic AI ecosystem. Increasing the output variability of production lines and processes almost overnight will be part of this.
Manufacturing, healthcare, transportation/logistics, education, hospitality, energy, defense, and healthcare are very ripe for this.
Humanoid Robots?
The flashy humanoid robot world will be less valuable economically and so much harder than anticipated.
Think about how complex/variable doing the laundry, the ironing, and putting the clothes away really is. Also, it's kind of egotistical to think that this amazing new technology should be applied to making things in our image - humanoids. Why?
Humans are walking error-generating machines that evolved with a few solid design elements like somewhat larger brains, stereoscopic vision, upright mobility, and opposable thumbs but not really optimal design for anything. Imagine if cows suddenly became super intelligent. Would they really want to make robot cows? If I had a magic wand and could robotize some living thing, I would probably merge an orangutan with an octopus with a condor. But I would really prefer to let opportunity and a long horizon drive design, not anthropomorphic tendencies. I would design for a box of LEGO approach not a least common denominator humanoid thing.
The NIVIDA News
Okay, so I am off track. Below is one great example, from NVIDIA, of this space leaping forward. It's a bit wonky with sentences like:
"We extract pseudo robot actions via a latent action model or an inverse dynamics model (IDM)." Well, of course they do....
The real point is this could be a 1000x improvement in a critical part of making robots functionally effective in various novel scenarios.
https://research.nvidia.com/labs/gear/dreamgen/
The Dark Side
Lots of people who have never managed a P&L or have almost no awareness of how shareholder capitalism (you know, that thing we call the economy) works have said over the past two years that the advances in AI are to make work better, to make workers smarter, to be co-this and co-that, not to replace jobs.
The simple fact is that AI and Robotics will increase productivity and outcomes in massive ways over the next three years. If I were thinking about hiring 100,000 programmers, analysts, HR people, marketing folks, entry-level lawyers, and factory workers in the next three years, I would be pausing that because I expect 40,000 to 80,000 of those to be unnecessary because of the throughput gains from this broad new tech. Find the HBR article written about WINS by way smarter than a robot, and think about the net job demand.
We are not prepared for that, and we are also not preparing.
End on a Positive Note
On a positive note I am optimistic that we will see a Revolution in Educational Affairs (yes stole that from Adm. Bill Owens) that can massively improve education outcomes and maybe do a much better job of matching future talent to future jobs.
This article, which I wrote in January, shows the scale of change in three simple graphs. Looking back at the change that happened makes sense, three years in advance ,that level of change was unbelievable. Three simple graphs that take 3 mins to fully grok.
The Real Distant Future 2029 -2032
The next three years will be a rocket ship of change. The three years after that will make those three years look tame. Four things drive that:
The beginning of broadly usable Quantum
Almost unbelievably huge practical impacts from Science: Material Science, Healthcare, Energy, Computing, Satellites/Space-based 'stuff' (e.g. LiDAR from space), Nano and multi-domain 'stuff' (e.g., nanobiotech, next-gen computational chemistry, fully functional biocellular models, robotic synbio, etc)
AI is 1000x better after normalizing for increases in demand, complexity, and scale (which means probably 1000000x better at a throughput level).
Networks as we think about them today cease to exist, and intelligent self-optimizing real-time fabrics are common, which include both connectivity and energy. Think of the difference between 2G and 5G or the changes in the Second Industrial Revolution, but landing in 2029 -2032 to sense the scale of change that is created.
Read that last bullet again. That means things in that fabric all become intelligent. Intelligent is a foggy word but that means proactive, pragmatic, preventative, personalized, pattern matched, predictive, process performant, precise, and permissioned. (Tony's10Ps). That is a different world and movement in the 10Ps direction hoovers up mountains of money in latent profit and cost pools.
Good thing geopolitics are really stable, otherwise this fast-approaching future could be really tricky.
