Thread by @jensenharris

It seems like every company is tripping over themselves in a rush to say their software is “powered by AI.”

But saying “powered by AI” is like saying you’re “powered by the internet” or “powered by computer code." By itself, it means nothing.

Here’s how I think about it:

1) Most “now powered by AI!” is just a rebranding of the same heuristics and rules engines software has used for decades.

When your email program keeps track of what folders you use most often and magically offers to file mails to them, that’s a simple algorithm, not “AI.”

Software has been adapting to users for decades.

The old Microsoft Office “smart” menus and toolbars rearranged based on your usage (“powered by AI!”) as early as 1999.

Learning spam and antivirus filters have been around for more than 20 years.

The advertised “AI-powered” climate control system in newer cars that turns down the heat as it approaches the target temperature is the same basic heuristic that was in the old late-night infomercial for the Ronco Showtime Rotisserie Oven.

“Set it... and forget it!” 🍗 🔥

But even earlier: 1952 Oldsmobile cars included auto dimming headlights that became less bright when it detected a passing car.

Though no software powered this feature, today it would be hailed as “AI.” Because AI has become, unfortunately, just an empty marketing term.

But it’s ok: Most software doesn’t need to be powered by extreme deep learning to have a great user experience. Rules and heuristics have a place. There is no shame in using them.

Heuristics are often smarter than ML at certain things because they are human-curated and designed.

An an example, Apple Maps uses some “AI” to decide whether it should show “Home” as a destination when I open it on my phone. It’s wrong like 30% of the time.

You know what would be better? A simple rule: show it when I’m away from my home, otherwise don’t. No neural net needed.

2) Even if a product is technically powered by machine learning, big deal.

Machine learning is widely available, cheaper every day, and easily applied to straightforward problems. Bragging that you use ML is like saying you use bubble sort or HTTP. It’s just table stakes.

In five years, most developers will be regularly applying basic machine learning techniques to solve problems.

It will be an expected skill, just like the ability to code using standard algorithms and data manipulation techniques today.

By focusing so much energy on this trivial aspect of the underlying technology used to make a product, we often forget to asses the real product questions:

“Does it work well?”
“Does it solve my problem?”
“Can I measure the results?”
“Does it make me happy to use it?”

Answering these questions is key, because there are both transcendentally amazing and absolutely horrible products that claim to be "powered by AI."

The AI-ness itself is mostly irrelevant; it doesn't have any intrinsic value.

Ask yourself: How does ML make this product better?

3) What machine learning does, though, is fundamentally make possible the creation of new categories of software.

As with GUI, the internet, and cloud computing before it, AI presents an opportunity for new, AI-native companies to create previously unimagined kinds of software.

The interesting space is not legacy companies “bolting on” AI to existing products, but instead totally new experiences that couldn’t have existed before.

At the intersection of conversational understanding + contextual learning + adaptive UI, so many new things are being born.

Cloud software eventually won out over legacy desktop software.

The same pattern will repeat here, as AI-native companies imagine and deliver entire end-to-end scenarios that wouldn’t have been possible by companies retrofitting legacy products with “new AI-powered features.”

But even here, machine learning itself isn’t the thing. It’s ML + heuristics + data + user experience together with smart product design and a problem that’s worth solving.

Companies which bring all of this together, harmoniously, in important domains, will rule the next decade.