Where Is the Machine Usefulness? The Faster Horse

Automation enables some of the tasks previously performed by labor to be produced by capital. https://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.33.2.3

Have we lost sight of what we want computers to do? Aren't they supposed to make our lives easier? Make them better? What we need is productivity, not job loss through automation. Incremental productivity over long periods of time is actually how we improve our quality of life. How do we do that with GenAI? And an aging workforce? And globalization morphing into... something else? (A crab, maybe.)

Anthropics "Computer Use"

Shouldn't "computer use" be called "computer computer use"? I mean, it's a computer using another computer.

A few days ago, Anthropic announced "Computer Use", the ability for their LLM to use a computer. It does this by taking screenshots and moving a mouse, and can thus accomplish things on that computer by itself.

I'm not going to explore what use looks like or how you do it, but rather ask if this is the best thing we can do with our time, the AI ecosystem in general, and where we're going to get the productivity from. Not just automation for automation's sake, but real productivity that gives us time back in our day and also gives us the ability to do the things we want to do, like making art instead of doing laundry.

We’re also introducing a groundbreaking new capability in public beta: computer use. Available today on the API, developers can direct Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking buttons, and typing text. Claude 3.5 Sonnet is the first frontier AI model to offer computer use in public beta. At this stage, it is still experimental—at times cumbersome and error-prone. We're releasing computer use early for feedback from developers, and expect the capability to improve rapidly over time.

It's kind of funny to me that this is called "computer use" because we're getting a kind of funny program (i.e. the Large Language Model–which of course runs on a computer somewhere) to use a computer...so isn't it really "computer computer use"?

Decades after their invention, computers are still hard to use. Sure, they can do a lot of things, a lot of new things, but they are hard to use. I hope you like clicking buttons and typing, because that's what you get! Even now, I'm still typing on a keyboard, writing text just like I did decades ago. Sure, it may be easier to check spelling and grammar, and I can easily publish a newsletter on the big "I" Internet, but overall it's still a pain, all of it. I'm cutting and pasting and clicking and adding links all by hand. Where's my computer jetpack?

Why Should a Computer Be Like a Human Being At All?

Why should a computer be anything like a human being? Are airplanes like birds, typewriters like pens, alphabets like mouths, cars like horses? ^[1:4]

Mark Weiser deals with this topic. I've discussed it in previous newsletters.

Weiser talks about how computers should be invisible, and also that computers are not like people. I would tend to agree with those statements, but the problem is that LLMs can't interact with computers simply through an Application Programming Interface (API), because most of what we do with computers is not through an API, but rather by clicking and typing our way through various applications to accomplish something. There is no way that an LLM could use an API to, say, research vacations and book flights and such. He would have to actually use a computer to do that. Also, in terms of visibility, we could be talking to the LLM and it could be using a computer behind the scenes and be invisible.

The Computer Revolution Didn't Deliver Productivity

Daron Acemoglu, winner of this year's Nobel Prize in economics, writes in the New York Times about the challenges we face in the near future, including AI.

...if handled correctly, these challenges [an aging population, the rise of artificial intelligence and the rewiring of the global economy] could remake work and deliver much higher productivity, wages and opportunities — something the computer revolution promised and never fulfilled. If we mismanage the moment, they could make good, well-paying jobs scarcer and the economy less dynamic. Our decisions over the next five to 10 years will determine which path we take. - https://www.nytimes.com/2024/10/17/opinion/economy-us-aging-work-force-ai.html

He continues:

A.I. is an information technology. It will not make your cake or mow your lawn. Nor will it take over the running of companies or scientific inquiry. Rather, it can automate a range of cognitive tasks that are typically performed in offices or in front of a computer. It can also provide better information to human decision makers — perhaps one day, much better. - https://www.nytimes.com/2024/10/17/opinion/economy-us-aging-work-force-ai.html

Is what we do in front of a computer so valuable? What we need is to get away from the idea that what we do in front of a computer is our worth. As we can see with something like "computer use," it's likely that the work can be automated.
It's what we do away from the computer that's important, which is hard to understand because we, or at least office workers or knowledge workers or whatever, do it. That's where a lot of us see our value.

Now, some of the things that we would like to do, for example, maybe writing as an art form, as something that one might be driven to do, can still be done in front of a computer or maybe a typewriter, but that is not the same as going into an office and sending emails and using various software applications. This kind of work seems increasingly likely to be done by an LLM - remember we are only at the very beginning of this LLM journey, what they can do today is probably the worst they will ever be able to do.

We Need Productivity, Not Just Software Automation and Displacement

Acemoglu argues that with an aging population, we need to automate with advanced machinery, and that is how countries like South Korea and Germany have done well, though much work remains.

Over the past three decades, Japan, Germany and South Korea have aged almost twice as rapidly as the United States is aging right now, which means we have models to follow. The good news is that their economies have not grown more slowly than those of other industrialized nations and that several of their labor-dependent sectors, including cars, machine tools and chemicals, haven’t suffered. The reason is simple: They introduced new machinery, including industrial robots and other automation technologies, to take over the tasks younger employees would have performed. - https://www.nytimes.com/2024/10/17/opinion/economy-us-aging-work-force-ai.html

As well, he goes on to discuss how our people, our workforce, is also not ready for this kind of change. Strangely, we may need to be MORE technical as a workforce.

Today too we are witnessing a period of rapid automation. The jobs of production workers are being disrupted with the rise of industrial robots and other automated machinery (Graetz and Michaels 2018; Acemoglu and Restrepo 2018b), while white-collar workers in
accounting, sales, logistics, trading, and some managerial occupations are seeing some of the tasks they used to perform being replaced by specialized software and artificial intelligence... - https://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.33.2.3

We need to find the new tasks that machine and software and AI can't accomplish. The new work.

Although software and computers have replaced labor in some white-collar
tasks, they have simultaneously created many new tasks...we show that about half of employment growth over 1980–2015 took place in occupations in which job titles or tasks performed by workers changed. - https://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.33.2.3

Advances in technology give us new things to do, possibly better things, but we need to be able to deal with the change.

The Faster Horse

Is a computer using a computer just a faster horse? We don't need a faster horse, we need a completely new method of transportation, one that our workforce understands how to build, operate, and travel on.

Further Reading

Link - https://www.nytimes.com/2024/10/17/opinion/economy-us-aging-work-force-ai.html

Link - https://news.mit.edu/2023/who-will-benefit-ai-machine-usefulness-0929

Link - https://www.aeaweb.org/articles?id=10.1257/jep.33.2.3

Link - https://www.anthropic.com/news/developing-computer-use

Link - https://os-world.github.io/