Mastery over information

A Paper by Jason Crawford

I’ve already described why AI has the potential to create a new economic era. Let’s consider what this might look like.

In the bold, ambitious future, AI workers may have enough different skillsets to enable entire virtual businesses. Today, if you have an idea for a software application, you might have to hire engineers to develop it, product managers to write the specifications, designers to create the interface, marketers to launch the product, and sales representatives to close deals—presuming you don’t have all of those skills yourself. To hire those people takes millions of dollars, so you have to figure out how to network your way to VCs and pitch them—presuming you aren’t independently very wealthy. This process takes years.²⁰ In the future, you might be able to spin up virtual workers for all the above functions on demand. Instead of 2 years and $2 million to launch a simple app, it might take 2 months and $20k. This would be accessible to many more people, and for many more applications. Many custom applications might be created that don’t even rise to the level of businesses, to run every reading group, little league, and knitting club.

Professional services might be democratized. Today only the wealthy can afford the best lawyers or doctors, or an accountant to do their taxes, or a therapist to listen to their troubles. In the future, first-rate services might be available to almost everyone—just as the average person now consumes fruits, spices, or clothing that once were the prerogative of royalty.

Another luxury service the wealthy can afford is private tutoring. In the bold, ambitious future, every student might receive personalized instruction. Teaching could follow the student’s interests, on a daily basis, to optimize for motivation. When a student is struggling, the lesson could be slowed down, simplified, or repeated as necessary for mastery; when a student is excelling, the lesson could be sped up or taken to the next level. Every student might be able to achieve their maximum potential.

A very small team, or even one person working alone, might be able to create an entire feature-length film, for a fraction of a percent of the budget of today’s blockbusters. This could unleash a flood of creativity, make an end run around the traditional media gatekeepers, and liberate us from the Hollywood rut of sequels, franchises, and remakes.

AI matchmakers might help us make all kinds of connections, both personal and professional introducing us to people we’d never otherwise have met, and triaging out the introductory meetings that would just be a mutual waste of time. And at the other end of the pipeline, AI might help us close uncomfortable ads—buying a car, selling a house, negotiating a job offer or a VC term sheet—saving us from processes that most of us find uncomfortable, and leveling a playing field that is currently tilted in favor of the more experienced party in each of these deals.

Language might no longer be a barrier, anywhere, for any reason. Imagine knowing every language in the world. Imagine being able to read any book, article, paper, essay, or blog post, no matter what language it was written in. Imagine being able to listen to any talk, podcast, or news report, with the “Babel fish” from the Hitchhiker’s Guide as your interpreter. Imagine this ability granted to everyone on Earth, beginning to dissolve the barriers between peoples and cultures.

Mathematics might soon be, in essence, completed: all open questions answered, from the Riemann hypothesis to the twin prime conjecture to the P vs. NP question.²¹ Unlike other branches of science, this would require no laboratory experiments, just a lot of computation; and the results could be verified via formal proofs—no need to trust the AI’s intuition or worry about hallucinations.

All of the above is based only on AI that runs in the cloud, or on your phone or laptop. A new generation of robotics, leveraging deep learning techniques and modern language and vision models, might dramatically expand what physical work can be automated.²² The obvious possibilities, already thoroughly envisioned by science fiction, need not be rehearsed here.

Also, all of the above is envisioning only our current screen-and-keyboard interfaces to software. Already AI is being incorporated into wearable electronics, such as voice-activated pendants or camera-enabled Ray Bans.²³ An AI that hears what you hear, sees what you see, and is always available for a conversation might be able to assist you better than any butler or servant.

Or, someday, we might perfect brain-computer interfaces and connect to the machines directly. Already experimental Neuralink devices have allowed a paralyzed man to control a computer; other devices are being tested to restore sight to the blind.²⁴ Information tools, from cuneiform to smartphones, have always acted as extensions of our minds, and the more ubiquitous and accessible they are, the more they can augment our cognitive abilities. The logical end of that progression is to merge with the machines.