Artificial Intelligence For Innovation

How Artificial Intelligence systems that create images from natural languages can propel innovation

A Doer's Manifesto

I am a big fan of Luis Perez-Breva′s "Innovating. A Doer′s Manifesto" . So big that I have two copies of his book, one in English and one in Spanish, just to make sure I didn′t miss anything in translation. I even took his MIT Professional Education Course .

What I find most attractive of Perez-Breva′s approach to innovation is his concept of a "hunch".

When faced to innovation challenges, many of us feel daunted with the process of being creative and coming up with good and original ideas.

Perez-Breva urges us to liberate ourselves from the pressure of having to think of something new. Start instead with a "hunch", a simple intuition about something that seems to be not quite right. Something that could be improved, something that poses a problem.

We usually consider innovation as the act of proposing an idea (the innovation) that will hopefully solve a problem. In fact innovating is a verb, the act of addressing a problem to hopefully find a solution.

But, how and where do we find those hunches?. Perez-Breva suggests to find inspiration in Science Fiction, in negating common assumptions and on daring to make preposterous combinations.

A preposterous combination consists in making questions of at least two things that are not supposed to go together - until now. The combination of a camera and a phone would have been preposperous for most of us in the 80's, for example.

The power of a preposterous combination does not follow from the combination by itself. I comes from work that comes after to try to make the prepostereous combination tangible. And here is were AI (Artificial Intelligence) can lend us a hand.

Artificial Intelligence for Innovation

In the last months we have seen how Artificial Intelligence systems that create images from natural languaje such as Stable Diffussion or DALL-E are becoming widely available.

The single description "A minion and his girlfriend at a beach in bahamas", renders the following image in seconds:

It is easy then to replace those 3-words-sentence sticky notes used in design thinking and brainstorm sessions for a more visual and tangible representation of the idea.

But these tools can also help us discover combinations among disconnected concepts, reduce creative block and stimulate out-of-the-box thinking. I generated 9 DreamStudio images for "an antigravity flying bicycle in the sky":

We get a gallery of images that can help us making the concept of an "antigravity bicycle" tangible and look into different directions. A slingshot bicycle and a ramp seem to be behind figures 5 and 6. Figures 2, 3 & 7 suggest enquiring into a bicycle with wings. Figure 8 looks like a bicycle with propellers, figure 9 with a balloon. I honestly can′t understand figure 4 (a bicycle-drone?) but figure 1 invites thinking of an external source of motion magnetic, antigravity, ...

Innovating by Tinkering and Playing with Artificial Intelligence

You don′t even need to start up with a fancy combination of words. Let′s say you start with "Clockwork Orange"

The results are not surprising since "A Clockwork Orange " is a popular icon and there must be thousands of related images in the AI database. But you can tinker with words to get other results. How about a replacing oranges for other fruits?:

Or adding more words like: "Clockwork Banana On Steroids":

How about keeping the oranges but tweaking the "Clockwork" into "Antwork"?:

Have your ever wondered what an "anesthesiologist flight attendants" would look like?. Pretty much like an group of anesthesiologists in a plane:

How would you describe a "solar energy harvesting shoe"?

How about a "shoe with iwatch features"?:

Conclusions

None of the above images is an innovation. They are only the first steps from a hunch which helps us to make it more tangible and reduce the creativity burden. They may take our mind into avenues that we might had not considered at the start.

This is what got for a "dresser and a washing machine":

Quite unimpressibly the system returned a dresser and a washing machine. But it also returned this one:

which may lead you to think of the dresser doing (somehow) the washing. You feed the AI system with descriptions and the systems returns you some images for you to work upon and move one step forward.

In the end these Artificial Intelligence systems are based in millions of images which already exist in the world (even if only out of the mind of someone). But that is right how innovation works, by combining existing parts in new ways.

As Perez-Breva likes to say, "Nothing is new at the outset of what we only later celebrate as innovation".

Final Note:

I would like to thank Javi López for publicly sharing his work with AI generated images, which inspired this preposterous connection between him and Luis Perez-Breva. The featured image of this post "Cthultu at the MIT campus in a dark rainy night" is a tribute to his ilustrated novel and book (which I am about to purchase for myself).