Natural language, technology, and intimations of AI

Consider the following bit of text:

^(?=.*[a-zA-Z])(?=.*\d)(?=.*[!@#$%^&*()_+\-=\[\]{};':"\\|,.<>\/?])[\w!@#$%^&*()_+\-=\[\]{};':"\\|,.<>\/?]{16,}$

If you're a programmer, you probably recognize that as being a regular expression. If you're not a programmer you probably don't think that looks too regular. Regular expressions look intimidating, and to some extent they are, especially if you're not a programmer. But regular expressions allow computer to parse text in much the same way you would.

Consider the following natural language prompt:

Minimum 16 characters, at least one letter case insensitive, two numbers, and one special character

A ten year old should be able to read that prompt, parse its meaning and come up with a text string that conforms to its criteria:

12&abcdefghijk

But in order for a computer to do that, it needs a regular expression. In fact, it needs the regular expression shown at the top of this post:

^(?=.*[a-zA-Z])(?=.*\d)(?=.*[!@#$%^&*()_+\-=\[\]{};':"\\|,.<>\/?])[\w!@#$%^&*()_+\-=\[\]{};':"\\|,.<>\/?]{16,}$

What if you were able to write a natural language statement, such as "Minimum 16 characters, at least one letter case insensitive, two numbers, and one special characters" and have a computer output a text string that conforms to the constraints listed in the natural language statement?

Well, that would be pretty impressive, and it turns out that someone has built it:

The person who created this tool is in high school. So consider: powerful AI technologies are allowing bright high school students to build powerful tools that make technology more accessible to more people.

Maybe regular expressions don't interest you. Well, check out Elicit.org, which allows you to use natural language queries to find research papers. Interested in the prevalence of HIV in Ashkenazi Jews? Here's a list of papers for you to check out:

These examples are but two of a rapidly growing number of tools which allow ordinary users to combine natural language with powerful AI technology. No longer do you have to be a programmer to use AI. You can just use natural language to do what you want.

I've shared these two examples of AI tech because they create a friendly interface for non-technical people to experiment with AI. More pertinently, these tools, and others like them, augur a much different future over the next decades than our present, given the rate at which the underlying AI technology is improving.

And, indeed, a lot of people speculate that we're in the beginning stages of an AI-powered productivity boom. Here's Holden Karnofsky writing about an incipient AI-led productivity boom:

Standard economic growth models imply that any technology that could fully automate innovation would cause an "economic singularity": productivity going to infinity this century. This is because it would create a powerful feedback loop: more resources -> more ideas and innovation -> more resources -> more ideas and innovation...

...

The feedback loop could come bac if some other technology restored the "more resources -> more ideas dynamic. One such technology could be the right kind of AI: what I call PASTA, or Process for Automating Scientific and Technological Advancement.

Holden goes on to argue, in the rest of his post that his PASTA conception will become a reality this century. It is part of a larger series of pieces that he has written, in which he argues that ours is the most important century in the history of humanity, due to how advanced AI systems will change humanity's trajectory.

All very heady stuff.