Computers: 1997 and Today

Janine Benyus wrote her book on “innovation inspired from nature”, Biomimicry, in 1997. The year I graduated. At one point, she lists all the things computers couldn’t do at that time:

1)    “Instead of typing into them, we would simply show them things…”

Today, your phone can be unlocked by looking at it. You show something that interests you to Google Lens and it can do a reasonably decent job of pulling up info about it.

2)   “…or they would notice for themselves.”

The lights in smart homes can turn on when you step in.

3)   “They would be able to answer not just yes or no…”

Siri. Alexa. OK Google. We’re spolit for choices.

4)   “Spotting someone who looks familiar, they would venture a fuzzy guess as to the person’s name…”

Facebook does it so well it’s almost creepy. Be careful what you wish for…

5)   “As they got older, they’d get wiser.”

This is debatable even for humans. But sure, as computers get even more data, they certainly get better.

6)   “Our computers, unlike our individualized brains, can’t learn to learn.”

It’s so commonplace today, we have a name for it: “machine learning”.

7)   “Redundancy”: Brains have it, computers didn’t.

When was the last time Google went down? Or Facebook? Or WhatsApp? Redundancy is built into almost every computer related field.

8)   Life evolves via change whereas “Computers can’t brook too much change.”

That was true until machine learning and Big Data. Today, computers process the data, find patterns, and change those patterns if the data changes.

 

So while so many things that seemed inconceivable when I graduated are commonplace today, it’s also true that computers/software have incorporated some of the how-to-work points that Benyus advocated in her book:

1)    “Computer, assemble thyself”

Ok, that didn’t exactly happen. But we let the algorithms evolve themselves. Feed the system a million picture of cats and the system will learn to recognize cats.

2)   “Survival of the Fittest Code”

This is the governing principle of machine learning today.

3)   “Giving up control”

We didn’t write algorithms to tell a machine how to recognize faces. Rather, we gave it generic instructions, and let the machine learn on its own using that + all the data from the real world being fed to it.

 

Michael Conrad lamented at the time the book was written that:

“The last thing the world needs is another new device.”

That’s hilarious, right? Today, everyone wants that new smartphone. Which, as my daughter says, allows everyone to do at least one Harry Potter spell. Harry needed his wand and say “Lumos!” to lighten up the room. Anyone with a smartphone can do the same by turning on the Flash Light app.

 

So yes, the smartphone is like a magic wand. And it would have been impossible to conceive of as lieing in almost every human’s hand when I was graduating. Just goes to show that predicting what can’t be done is very dicey.