From the article:

"Last year, a team of engineers from Barilla pasta came to see if they could reproduce her technique with a machine. They couldn’t"

I expect it certainly can be automated, but there's an ROI to be considered. If it was just a matter of new dies and some process tweaks using existing machinery, it absolutely would be done and that's likely what the Barilla people were trying for.

If new technology has to be developed to match the manual process, that's far a riskier proposition just for the purpose of adding yet another box of pasta in fancy grocery stores.

It probably makes far more sense, economically, for a small number of elite chefs to learn and replicate the technique for their restaurants. I am sure folks are trying.

She's measuring elasticity and water content through feel. Then she applied the correct amount of salted water or regular water.

My assumption was they didn't have an existing way to measure elasticity / water content on pasta that thin economically.

To your point, new tech == tell professors who might care and move on.

> I expect it certainly can be automated

Why would you expect that? Many processes are resistant to automaton.

I would be surprised if it could be automated. The amount of nerves and muscles in the hands is enormous, and many of the materials have no artificial equivalents. Cameras don't even get you close because you need to see inside the dough to make measurements. And quite a lot of the process isn't observed at all, it's modeled in the chef's head and hands. Things like the ratio of muscles around her fingers encode reactions to data coming in from nerves. Very difficult to study let alone replicate. A process like hand pulled pasta seems like the absolute worst case for automation.

I'm not suggesting it would be easy, but there are many manufacturing processes that routinely do things which are staggeringly difficult and sensitive in all kinds of domains. It really is a matter of understanding the process and compensating for the things a machine can't do in the same way as a human.

I posit that the ROI for such an endeavor would be disappointing in that all you will have is another type of fancy pasta. That would sort of take the magic and cache out of the dish as well.

As I wrote here in another comment: I expect conscious AI to be achieved sooner than replicating what human hands can do.

My bad. For some reason I missed that part.

When reading it, I thought about another HN post in which a Japanese classified cucumbers using deep learning and thought that maybe that could be applied too to get the right texture and moisture.

It seems to me, but the main question is... would it be useful?

It depends. If people like it or if it gets a status of "haute cuisine", it would be useful (in terms of money) and would solve the problem of people not wanting to do it because of its tediousness.