[MUSIC] >> Hey everyone, my fellow AI nerds.

So I'm David Cole and I work for Colgate-Palmolive.

I know this is an incredibly interesting company.

This big tech company, just kidding, our technology is mostly chemical.

But we've been using it in a really interesting way and I just want to talk to you guys about that a little bit.

So the level set, I'm also not an AI visionary.

I'm definitely not a developer, I'm not in marketing.

In fact, marketing is like my frenemy.

We work with them a lot, but I find them to be a little bit difficult.

But that's the creative side of me.

I am a creative thinker.

I like to think that I'm a decent designer, and I'm a very curious person as well.

Last year, here we go.

Last year, our buddy Tim, who spoke earlier, amazing presenter.

Thank you, Tim.

I can't wait to tell my company that I did therapy so that I can be nicer to the AI to get better results.

Tim had said, "The best thing you can do in AI right now is to ask questions, tinker, and don't trust anyone that sounds too confident."

This super resonated with me because I had only begun with my team to just be on that AI journey that everyone was on.

We were using DALI, we were using whatever version of GPT that was out, I guess three, whatever.

We were making some mistakes.

I think we all went into it with this expectation that it was going to solve all of our problems, that the robot overlords were going to replace us.

And it's just not true.

Gen AI is a predictive pattern tool.

I had to write, I'm pretty sure, not 100 percent, but I feel like I was validated earlier today by Tim and Noah, so thank you for that.

It's not a truth tool, as we saw, kind of demonstrated earlier by the math.

It's not a divining rod either, and it's not a magical wand.

But what it is, is an excellent partner for creativity.

We are not going to be replaced, we think, we hope.

I'm confident in that.

It's not a deus ex machina.

I think that's really, really important.

It is not a god sent from the heavens to just unravel and solve the world's most difficult problems.

Get a little cerebral for a moment.

For those of you who don't know, deus ex machina actually is a Latin term that translates from god to the machine.

It was a physical device used by the ancient Greek and Roman people who did the dramas, where they would physically crane and lower in a god for any problem that got too tangled or difficult to solve with the 15 minutes they had left, which will be a very similar thing that will happen in my presentation later.

But in modern usage, clearly it means it's the device in literature where you get a little lazy and you're like, "I need to solve this."

Like, insert magic, you know?

So what we're saying, basically, is the world is never going to be as slow as it is right now, today.

So let's totally adapt.

But I think it's important to recognize that DNA is not here to just conveniently solve everything and make our lives super, super easy.

But what it can do is make our lives better, make us more superpowered.

So I want to show you guys a quick project that we had done at Colgate Palmolive.

I'm not going to get too specific because the legal team was very precious about what I could and could not show.

So we're going to call it Project Marimba.

Project Marimba was really, really cool.

We work with a lot of scientists.

We have a lot of cool technology in the form of chemistry that we can share and do and different things with.

And but often they don't have a very solid value proposition or a piece of packaging that makes sense for the chemistry.

So the opportunity was this.

Let's make an all-purpose cleaner more sustainable through a new concentrated formula.

That's awesome.

That means less plastic and less shipping costs, weights, that's good for the planet, that's good for business.

We want all of these things.

But that's generally the opportunity.

But our briefs are usually pretty bad and they're like, it's for this brand.

So it was like early days, our Deus Ex Machina, we're like, let's let Dolly solve this for us.

So we were like, design a cleaner bottle for the XXX brand.

We won't tell you which brand.

It should be a concentrated product, be small in scale, have an easy to use cap and have structural ribs.

And the very first hallucination that it spit out is this very disgusting purple bottle with actual herbs in it.

So a couple of things here.

One, it showcases that it's not going to solve your problems.

It also showed that we didn't really understand prompting yet, but we'll get it.

We're not going to go too far into that.

We definitely didn't understand it.

The Deus Ex Machina of it all.

So in all the early experiments, we were like purple bottle.

We sprinkled in some words like joyful cleaner, by the way, it's cleaner.

And it yielded wildly unpredictable results.

They were none of them were very good, clearly, and none of them were particularly useful.

We were really proud of ourselves for like trying.

We're like, oh, bad on the back.

But in the business world, unfortunately, that doesn't really get you anywhere.

So we started to break it down.

So what is even a bottle?

You know, the marketing team seems always like we need a bottle.

I'm like, cool.

Why?

What's it for?

So we just kind of generally broke it down.

All right.

What do we have to work with?

We have this yellow cap.

We've got tapered shoulders.

We've got structural ribs, label space, blah, blah, blah.

You guys know what a bottle is.

But we have all this other great stuff that we don't think about enough or very often to make it intentional, intentionally designed.

So we have, you know, the brand values, the attributes, the tone of voice, et cetera, et cetera.

But even that, I find, isn't like a super solid brief.

It's like design a bottle, design it with these attributes.

I really want to be super intentional about the opportunity for design.

So let's find this meaningful why.

Like, why should this exist?

We don't just sell.

Well, I think the business probably just wants to sell like a ton more stuff and like, you know, margin, profits go up, stock market, blah, blah, blah, blah.

But for me as a designer, it's really important to have a meaningful why.

And so we started experimenting with Gen AI, different models to help find that.

We use Miro quite a bit internally.

I think everybody hopefully knows what Miro is if you haven't.

Great whiteboard tool.

But they've got a set of beta pieces inside of it that my favorite one being the mind mapping.

And you can ask it to generate all kinds of stuff.

So you can start with helping to align on strategic questions and you can start with one that's like maybe OK.

And instead of sitting in a room for hours with your team coming up with these different topics to think about, you can just have it quickly populated.

You're skimming all the best ones off the top.

It does ideas, does all this great stuff.

We're taking a lot of the stuff we're building out in Miro super quickly.

We're popping them over into our good friend, ChatGPT, and we're actually using Gen AI to help inform the discussion guides that talk with real people.

That kind of exercise, sadly, is one of the things that takes a lot of time.

It takes a lot of thinking.

And we're actually really able to expedite that kind of work so that we can be more curious more quickly.

So we're using it to summarize all these ideas or all of our one to one interviews.

So we can now do twice as many interviews as we used to be, be way more people centered, technically, I think, and pump out all the summarizations to get to these most important insights like very, very quickly.

We're also feeding these insights that we're having it summarize.

Sorry, this is like a really long dialogue.

But to get to that concept framework.

So this is the meat and potatoes of all the work that we really want to be doing.

So how can we feed the insights of the ChatGPT to generate the concept framework?

But we're also using it to start to create our digital twins.

We'll actually take some of those summarizations, have it generalize some of the thinking.

We'll, you know, lowest common denominator, 20 people into three or four.

Give them a name, give them a narrative, a story.

They'll have a job if that's helpful for whatever you're doing.

And we'll actually use them as part of the tools to brainstorm with our teams.

This kind of work is actually making us better designers because it's leading us to ask better, different questions, more curious questions.

So what started as just a bottle design for concentrated, which is like boring, is far more interesting when we're talking about things like, hey, how can we immediately communicate the waterlessness of a product?

Like what design decisions might you make in order to do that?

You know, how does a yellow cap become a beacon of authenticity if that's really important to the brand that you're working for?

Or how might the silhouette convey a sense of optimism beyond just like the semiotics of the bottle?

Like what is it doing in your life?

So using that kind of thinking, we're 10xing our workshop activities.

Again, going back to Miro, we've used all that rubric that we've started using, building out of chat.gpt, all the different information we gathered through our people's electricity exercises.

And we're speeding this stuff up as well.

So brainstorms are super, super fun.

I love having thousands of ideas.

Who doesn't?

The thing about brainstorms that no one really talks about is sucky, is summarizing everyone's ideas that are really close together, often pretty much the same idea, you know.

So we're actually just lassoing these bad boys.

We're summarizing them really quickly.

They're also sorting them for us, so like which ones are connected.

And we're saving heaps of time on that front, which is like incredible.

It's such, such good efficiency for us.

And we're spending more time thinking about the problem, which is like big thumbs up.

Additionally, when you're working with different teams, whether you're working with, I'm working with scientists or marketers, those guys, we want to be able to turn their words, which are absolute vapor, into something meaningful.

So we're democratizing the visualization of that using these tools as well, allowing more participants to like be more involved and be more invested in their ideas and thinking through the ideas that they're having.

So this is what having the, using all that can start to look like, a good old fashioned idea having.

So top left corner, how can we immediately communicate the waterlessness of a product?

I don't think this is the kind of stuff we want to hand over to our computer overlords, guys.

This is the kind of stuff we want to be doing ourselves.

Good idea.

Hey, let's just put a big hole in it.

That communicates waterlessness.

Immediately can see that there's less water in it.

You know, how might this silhouette convey a sense of optimism?

You know, this brand happened to be have a like a really strong fragrance that helps extend this idea of clean.

Hey, maybe we just turn the whole bottle into a diffuser.

You know, hey, maybe that diffuser, we don't want it just sitting out.

Hey, what if the bottle was a doorstop?

You know, we're having all kinds of wild ideas.

All of this is really cool, and I don't think it should be replaced.

But once you start to make decisions around these things and start to get into it, you need to get to the nuts and bolts of executional design.

And we're also using Gen AI to help us further that along.

So the devil in those details.

So instead of asking it to make the whole bottle like you guys saw previously, we're asking ChatGPT or Dolly, the different parts of the brand, what does authenticity look like?

What does joy and optimism look like?

And we're starting to get these really interesting kind of textural elements.

But we then need to interpret some of these.

So we direct prompt that and be like, all right, thanks for the blobs.

Can you apply these blobs to some bottles so that we can really understand what it looks like?

So we start to do that.

And then we go, oh, excuse me.

This is how we kind of go, sorry, I'm running out of time.

So all of a sudden I'm hustling.

I think it's really important to move from that inspiration to execution.

Like, go ahead, let's let the machines do a lot of the heavy lifting.

But we still need to do the old fashioned sketching, building, et cetera.

Like, you are not going to get ChatGPT to make a tight sketch so that you can start to inform your CAD model.

Like, this needs to be volumetrically sound.

It's just not going to happen, at least not yet.

Maybe we're in some cool future.

They'll probably be in like a week and a half when they do that for us.

But we're not there yet.

So thankfully employed.

And then, of course, concept refinement.

We do a lot of like engineering validation, supply chain, blow molding, et cetera.

So yeah, yes.

Don't worry about that, legal team.

So all this is to say we're trying to look at using these tools in really cool, in different ways that aren't just like trying to kick out like the solution, because it's just not how these things are end up working.

What we wanted to do is remove assumptions, leading to more creativity, more diversity of thinking to go further, much faster.

So I'm going to kick it off or end it with, hey, yeah, it's helping us like improve efficiency and reduce costs, but more importantly to me and hopefully to you guys, that it's allowing us to become more people-centric, spend more time asking the questions that we want to ask, create more diverse ideas, more quickly supercharge those positive outcomes.

So I quoted Tim Hwang earlier today or early in the presentation.

If you guys get the opportunity to speak next year, please quote me.

Use Gen AI to turbo the task between curiosity and creation.

Thank you.

[ Music ]

‍