Arnold Kling  

Race Against the Machine Watch

PRINT
Clarity on Clearing Houses... The Great Stagnation for Colle...

Robin Hanson says,


We've seen many, many decades of steady progress in automation. We've seen what the rate looks like. We've seen that it's slow, methodical; it takes time to slowly chip away at the kinds of jobs humans do and get machines to do that. We're not on any sort of sudden threshold where suddenly a vast flood's coming over the gates. We are slowly chipping away and we'll continue to slowly chip away

It is an interesting discussion throughout. Robin is arguing that even though computers may be getting smarter exponentially, their economic effects are merely linear. Maybe. But a long S-curve can look like a straight line for quite a while when you are near the bottom, but then the curvature starts to matter.

The title of this post comes from an e-book that is on my list of things to read.


Comments and Sharing


CATEGORIES: Growth: Consequences



COMMENTS (6 to date)
jstults writes:

Possibly related, Boyd's Laws:

  • Logarithmic Law of Arithmurgy
    Insight grows logarithmically with the number of floating-point operations.
  • Extended Moore's Law
    Computational power and scientific data as measured by operations and bytes, both grow exponentially with time.
  • Corollary: Linearity of Progress with Time
    Knowledge, as opposed to mere facts or data, grows linearly with time.

Becky Hargrove writes:

Lifted from the discussion, Martin Ford says "You may be a very well rounded person but that's not what you get paid for". Robin responds "When you give Watson(of Jeopardy) a different game show he does not do so well." And then Martin Ford answers, "Most people don't do a different game show."

Well, people do "lot's of different game shows" but two things happen: 1) they do those different shows, monetarily speaking - generally at different times in their lives. And, 2) people do numerous game shows all the time, but for the most part just don't get paid for that life responsibility. The reason they don't get paid is that is this a different and equally important economic function that needs to be better understood, so that people can retain their economic and social freedoms. Plus, when people understand what the wealth creation functions of human skill actually are, computers will cease to be viewed as the enemy. They will become the 'help' that humankind can harness. As Robin said, "humans and machines complement each other".

I think that it is hard for Robin to see a tipping point in the present, in part for what that seems to imply, i.e. the idea that now somehow there needs to be some permanent redistribution from the government for the have-nots. Nothing can be further from the truth, because if government goes down that road there won't be enough wealth to maintain technology. Wealth creation lies at individual to individual levels as the ultimate compliment to money based technology. 'It is our interconnected economic value that machines can never replace.' (from my notes, I can't remember if I said that or if Robin did. Help Robin, was that your comment?) Machines are the guests in the world we create for ourselves.

Robin Hanson writes:

I think that within a century we will see a speedup and a big sudden effect. But we aren't close to that point now.

Joe Cushing writes:

If automation is set on an S curve, this is a solution and not a problem. It's a solution to millions of baby boomers who will be leaving the job market over the next 15 years. It's already starting and the mass exodus is accelerating. In a world where few people have to work, having means of capital ownership, such as 401ks and IRAs, spread over the population will be more important. You can buy shares of companies that use machines to do work in your retirement account and own their productivity. You can own the output of a machine instead of owning the output of you. Then you can sip martinis and take trips instead of working--for longer and longer periods of life. Already people do this for 20 years on old age and 20 years in young age. (The young part might be a bit different than drinks and travel but you get the idea.)

The problem is, people are born with $0. It is usually their labor that allows them to accumulate wealth. If there isn't enough work for everyone, wealth will have to be inherited. This gives much power to older generations and less freedom to younger generations. We are already seeing this as people live with their parents till they are 25.

All this said, the growth sector is growth itself. The last thing machines can't do well, is invent and construct better machines. Machines don't have creativity. There is evolutionary software though.

jb writes:

IMO, it's absolutely an S-curve, because machine-equivalent IQ is going up linearly, and IQ is on a bell curve, so the machine advancement is simply replacing the people at-or-below its equivalent on the IQ curve.

The 'tipping point' will be when we get into the 70s, 80s and 90s IQ equivalent. I don't know how long it will take to get there, but given things like Siri and automatic cow-milking robots and such, it surely can't be 100 years.


Mark Bahner writes:

Ray Kurzweil has a relevant observation that I'm too lazy to find.

He said that the number of Internet connections was doubling every two years from like 1980 onward. But people didn't notice until the 1990s when it was going from 10 million to 20 million to 40 million every two years.

That Apple 4s thing looks pretty slick. Imagine that 10 generations (in Apple time) into the future.

I can easily imagine in less than 30 years having all the food I take out of my refrigerator and freezer tracked by the refrigerator/freezer, and having all the replacement food delivered to my house by an automated vehicle. That eliminates grocery stores and all the people who staff them, and all the people who build grocery stores and all the refrigeration and freezing components inside them. It eliminates the parking lots at grocery stores. Electrical energy use is dramatically lowered.

Robin compares mechanization with computerization. But it's a whole new ballgame. The most unique thing about humans has always been the human mind.

Comments for this entry have been closed
Return to top