Axil Axil <> wrote:
> Might robots prove so cost efficient and reliable that restaurant
> employers replace a significant number of workers with these robots?

I say yes. I think in 10 or 20 years nearly every employee at every
restaurant will be replaced with a robot. In Japan automated sushi
restaurants are becoming popular. One of them opened in Atlanta recently.
Here is the web site and a short video advertisement showing how it works:

There is more automation here than the video shows. You do not see the
machines that make the rice and form it into sushi, something which until
now required a skilled sushi chef. (Supposedly, it did.)

Here is a British video showing more of the automation at this restaurant

I am not a fan of sushi, but this place is fun and I thought the (non-fish)
sushi was pretty good. My daughter says it is meh.

ANYWAY . . .

I find the video of the burger flipping robot thought-provoking. Here are
some of the thoughts it provokes:

This hardware is custom-designed to flip hamburgers, coupled with what I
assume is a general-purpose, vision-enabled robot, probably about as
powerful as the Baxter robot.

A person worried about the future of employment might feel sanguine looking
at this because it seems like it is a custom solution for one application.
You would have to design another robot to lay out the buns the way the
human worker does in this video. You would have to design yet another robot
to crack eggs, and another to make salad. It seems every job will require
specialized hardware and perhaps specialized robotics and software. You
might think that it would take a long time to make all the robots we need
to run a restaurant, not to mention all the robots we need to do housework,
stock grocery store shelves, or do carpentry.

I don't think so. I predict faster progress for two reasons. First, many
programmers and other technical experts will gain experience applying
general-purpose robots and they will soon learn how to apply them more
quickly. They will be more of these people. They will start their own
companies, branching out into new applications. They will train others, who
will train others. Second, the technology will become more general-purpose,
and less custom-designed. Even the parts which are custom-designed will be
easier to design. Once you know how to flip hamburgers, you can flip other
objects, or form pancakes, or wash lettuce.

The situation reminds me of software in the 1970s. In the 1960s, every
company had custom-designed accounting software made specifically for that
company, often in-house. In the 1970s specialized software firms began
offering packaged accounting software that could be customized for various
industry segments.

In the 1980s general-purpose tools  and database tools became available
that made it easier to write accounting programs. Computer memory and disks
expanded by orders of magnitude. Extremely powerful accounting program such
as Peachtree software became available. This was suitable for nearly all
small businesses, and it reduced the need for industry sector specific
solutions. Accounting software for large companies is now offered by a
small number of large vendors such as Oracle and SAP:

Few companies develop accounting programs in-house.

The other reason I predict faster progress is because people are developing
self-driving cars, which is one of the most difficult applications.
Tremendous progress has been made in this. Much more than I predicted a few
years ago. This is one of the most difficult jobs that can be done by a
robot. If it is done wrong, people will be killed and tremendous
liabilities will be incurred. It seems odd that corporations would start
with such a dangerous and critical application. You might think they would
start by making robots that fold sheets and flip hamburgers, or do other
jobs that cannot accidentally kill someone. They are concentrating on
self-driving cars because the market is gigantic, the potential profits are
gigantic, and the penalty for getting left behind by competition and rapid

There is no longer any question that the technology works and it is safer
than human-driven cars. All automakers are frantically developing this
because if they do not they will not survive another 20 years.

My guess is that once you develop robots that can drive safely, developing
robots to crack eggs or fold sheets will be easy in comparison. A whole
generation of robotic engineers and programmers will cut their teeth
learning how to make self-driving cars. They will soon be making other
things. This is similar to what happened when people developed the MIT
whirlwind project, Princeton's AIS computer, the IBM stretch computer and
the IBM 360. The young engineers and theorists who participated in these
projects went on to develop the computer industry in the 1960s. The AIS
project was directed by von Neumann himself. He had a lot more practical
knowledge and engineering ability than most other 20th century giants of
physics. If five or 10 people of his caliber become involved in developing
robotics for several years you can bet things will move quickly.

One of the first uses of the AIS computer was to develop the hydrogen bomb.
That was a rather complicated application compared to some of the ones that
followed. In other words, like self-driving cars, they started with a
difficult project. Because that's where the money was. That's what the
customer (Uncle Sam) wanted.

- Jed

Reply via email to