USPTO kicks innovation into overdrive with new AI patent guidance

On February 13, the US Patent and Trademark Office (USPTO) issued summary guidance on whether artificial intelligence (AI) can be named as an inventor.

The USPTO says yes. However, there’s a catch. Each filing to the office must include at least one (real) person who was “significantly” involved in the research pathway. What exactly the word “significantly” means in practice was left deliberately vague in the guidance.

In other words, the USPTO has just made a lot of extra work (and more money) for an army of lawyers on both sides of the patent fence. Without a bright-line test for when grains of sand become a pile of sand (that’s an analogy), it’s not clear what role humans will need to play in the process of invention in the years ahead.

That’s concerning, to say the least.

One assumes the USPTO now thinks it has solved the problem of who owns AI innovation. But the guidance it released has created a whole new set of problems that go much further than the USPTO’s remit and actually could become a serious societal risk.

Innovation (which is a sister concept of invention) is an intangible asset. Companies race to create unique products based on fresh techniques because doing so gives them an advantage in the market. If no one else can do what you do, that can be a huge “moat,” in business jargon. Investors look for a moat as a measure of the strength of a potential investment.

A successful invention can be highly lucrative, but being innovative isn’t easy. The path of research and development (R&D) is fraught with the risk that it will never produce anything of use and accountants often keep a peg attached to their noses every time they look at the expense sheets. R&D can be a huge drain on resources.

Until it’s not. At some point, a clever R&D project could end up creating world-changing technology. Maybe even be the cure for cancer. And all those sunk costs that felt like money walking out the door will instead have produced a valuable asset. In other words, R&D might just be the most important intangible asset a company can have.

So, what’s interesting about the recent USPTO guidance is the consequences it might have on R&D. Not necessarily on the quantity of R&D produced by companies. After all, super-powerful AI machines that can iterate at lightning speeds will probably mean far more ideas and products will appear on the market in years to come.

Rather, the consequences will be on the humans since the guidance makes one thing crystal clear: the minimum number of people required for a patent, in collaboration with AI, is one.

Not two. Not 20. Not even a whole team of researchers. Just one single human who has “significantly” contributed needs to be on the patent application. It doesn’t take much to guess what every company is now thinking. Why bother hiring dozens of expensive engineers?

In fact, all that staffing money is far better spent on buying more GPUs for the AI. After all, AI doesn’t get tired, doesn’t ask for salary raises and can be constantly improved with more data. Not even the smartest, hardest working or most absorbent person can compete with this. It makes perfect sense to invest in AI machines while spending zero on hiring more people.

This is the kind of disruption that will spiral out in unintended directions as AI replaces jobs.

Who needs a checkout teller when automated stations can do the same thing? Who needs factory workers when robots can work faster? Those at risk most in these situations were low-skilled workers. But no one seemed to care about them, so AI developers pushed on anyway.

Now we are starting to see who will truly be replaced by AI. It’s not only the low-skilled. It will also hit the Smartest People In The Room™, the very people who created AI in the first place.

If the future belongs to those who can create excellent human/AI teams, the people required for this future may turn out to be a far smaller number than the Smart™ people ever assumed.

Given the power of AI in 2024, imagine what might be coming next year. How about 10 years from now? It’s possible that one person could team up with an AI, feed it basic directives and produce the R&D output equivalent of entire multinationals.

It’s even possible that we could get companies valued at more than $US1 billion – with only one human on the payroll. Think about that.

Maybe the Smart™ people will each own an AI and set up an innovation company of their own. Perhaps hundreds of thousands of one-person companies collaborating with superhuman AI, each valued at unicorn levels, will pop up. Maybe everyone can use AI to become a billionaire innovator.

But that’s doubtful.

The money for investment must come from somewhere. Are investors more likely to support an AI project out of, say, Moderna or a one-person AI company based out of their garage?

You might say both investments are good bets, but you’d be discounting the importance of Moderna having a constellation of intangible assets that can’t be replicated in a garage.

For example, Moderna has deep relationships across the sector and excellent ties to supply chains that can source key items at a moment’s notice. Its leadership knows people. They can get innovation into the right rooms where the right decisions are being made. Moderna has a more holistic understanding of the pharmaceutical sector and can task its human/AI teams with developing the R&D that makes the most commercial sense.

A one-person AI company in a garage simply can’t replicate this constellation of intangible assets. Said differently, the future will be owned by companies with a bundle of intangible assets that figure out how to empower a handful of their staff to use powerful AI in innovative ways. One-person human/AI teams will exist, but they will be owned and operated by the big boys.

In other words, while the USPTO guidance is about to kick AI innovation into overdrive, very few people will be along for the ride. It’s hard to overstate the implications of this.