Human communication

Too many emails? Let your bot answer it

What will it be like when you have your own bot and it is as good or better than you at many daily tasks?

The answer may come sooner than you think. Google’s new Pathway Languages ​​Model, which is not yet open for public testing, is the latest advancement in artificial intelligence. The technical explanation is that the neural networks have been scaled to 540 billion parameters for “breakthrough performance”. The practical effect is that the AI ​​is now more adept at engaging in natural conversation, explaining new jokes, and writing code.

I expect that most written communications will eventually be done by robots. I could train my bot by letting it read all my previous emails and other writing. Eventually, my bot responded directly to most of my emails, though it might hold some aside to ask if they deserved a personal response.

It sounds convenient, and in many ways it will be. I will have more time to take walks and read books. But think about the larger balance. If more emails are read by bots, then more emails will be written by bots. Of course it already is, but in this new world emails composed by bots will be at least as good as human emails, and at least as good at passing through filters as I have put in place to protect my time and attention.

A kind of arms race will ensue. Overall, I expect the number of quality messages and emails to increase. Woe to those who don’t have a very good filtering bot.

Imagine negotiating or discussing terms in such a world. I might receive a proposal from your bot. Is this a real and legally binding offer? Or is it just a ruse to trick me into revealing information about my trading strategy? In some cases, bots can handle these issues smoothly and present both parties with a final settlement. In other cases, negotiators might insist on a face-to-face meeting, both to know they’re getting “the real deal” and to limit the potential back-and-forth. For some real interactions, online written communications will no longer suffice.

Think of the college admissions essay, for example. These days it is important. But if the bots get good at writing, candidates may have to go for a personal interview instead. Countermeasures could then evolve. Maybe there aren’t enough admissions officers to conduct all of these interviews. So why not let the candidates spend two days together, record all the procedures and let the bots issue notes? They could even measure who told the most original jokes.

In this new world, writing skills will matter a lot less, and personal charisma a lot more. This is not necessarily a positive development. It will be more difficult to use handwriting as a measure of a broader skill or intelligence.

If you are single, bots can change your use of dating services. It seems tedious to have to swipe left or right all the time – and besides, do you really trust your own judgement? Instead, you can let your bot choose for you. If you told him you were interested in a potential mate, he might even send you pictures of what your kids might look like.

As with email, however, there are some potential complications. The good news is that your bot can quickly sort through the set of available candidates. Maybe your soulmate was 3,472 hits – and maybe now, thanks to your bot, you find that person instead of giving up. The bad news is that the most datable and marriageable candidates could be removed from the eligible pool. Liquidity in the dating market could dry up for less desirable candidates. Tinder hides from you the reality that a candidate might only be a 4 on a scale of 1 to 10, but not your bot.

Artificial intelligence brings great benefits to a wide variety of tasks, such as detecting payment fraud, improving medical diagnostics, and sending rockets into space. The potential problems arise when AI systems interact directly with human attention – and the activity involved involves a lot of matching and filtering. In these cases, advances in AI may exceed our human abilities to participate in the process. And maybe there isn’t a final stage, at least not anytime soon, where we can fully rely on AI.

In the meantime: If you have any questions about this column, you’ll have to count on me for the answers. But maybe not for very long.

Related to Bloomberg Opinion:

• The high ideals of Google AI Unit are tainted with secrecy: Parmy Olson

• Confronting the potential of AI to create new chemical weapons: Lisa Jarvis

• Go ahead: ignore this email: Stephen L. Carter

This column does not necessarily reflect the opinion of the Editorial Board or of Bloomberg LP and its owners.

Tyler Cowen is a Bloomberg Opinion columnist. He is a professor of economics at George Mason University and writes for the Marginal Revolution blog. His books include “Big Business: A Love Letter to an American Anti-Hero”.

More stories like this are available at bloomberg.com/opinion