Design + AI = Good or Bad?

It's a very full room.

I was, secretly hoping like five people would turn up.

So thank you all for coming.

Really appreciate it.

[video plays] We got what you want, we got what you want, we got what you want, so come and get it.

We got what you want, we got what you want, we got what you want, so come and get it.

How was it?

It's so exciting!

Putting this on ABC free to air, 'Juicy' We also begin today by acknowledging the traditional owners on the lands in which this conference is being held, the Gadigal people of the Eora Nation, and we pay our respects to their Elders past, present, and emerging.

We extend that respect to all Aboriginal and Torres Strait Islander peoples here today.

We recognize their continued connection to the lands and waterways in which we get to live.

And acknowledge that they never ceded sovereignty.

They are and always will be our original storytellers.

Hi everybody.

We have a question that's been swirling around in our collective minds.

As our worlds are getting busier, faster, more rapid, where, as designers, product managers, and engineers, has our time for reflection gone?

Are the designs and products we are creating contributing to a good world or a bad world?

So to us, our role as designers is about bringing the human experience to the forefront.

And now more than ever before, I think you might agree that role has never been more important as the development of AI and personalization explode, morph and integrate into our daily lives.

So what are we here to share today?

Today we're here to talk to you about the ABC and how we're designing at a time of rapid adoption and of personalization and AI.

We use autoplay as a bit of an example of rapid adoption and about how blindly just adopting a feature, no matter how good it is, might not always be the right choice.

And by showing you how in the ABC Kids space, we've actually reframed autoplay to be a better parent led experience.

We'll then discuss the adoption of AI in products.

We'll showcase some amazing principles that were developed by our ABC Audio teams, and then put into place in ABC iView, with the hope you might take these away and you can find them useful in your daily work.

Throughout this, we'll be asking ourselves the question, are we designing with enough time to understand the consequence?

By the end of today, I hope you leave with four ideas.

Adaption versus adoption of features, have some principles to guide you in your development, make time for critical thinking and for slow down, and to always consider the impact of what you're working on.

Not everybody here will have our lived experience of the ABC.

I grew up with ABC in my lounge room with Play School.

I also grew up with it being in my kitchen with my parents listening to the radio and in the car on the way to school.

I couldn't escape it.

And then it was a place that I sought refuge when I was a teenager through Triple J, and Recovery and Rage.

So it's been like a very prominent, brand and experience in my life.

But some people here may not have had that experience.

So why is the ABC?

Who is the ABC?

The ABC was created in 1932 with the goal to provide information, entertaining and educational services that reflect the breadth of our nation.

When we think about 1932, it was in between World War I and World War II.

I can't even imagine how scary that time was for people.

They had, they know what war looks like, and they were scared of what was coming.

And what was really interesting is there was lots of misinformation swelling around, like lots of ma and pa radio stations.

There wasn't a single source of truth.

And that is where the ABC stepped in.

We started off as a single radio station.

And then we turned into broadcast TV, and now, and today, our products are more than just TV and radio.

We have a large digital ecosystem, and the products we make help to share the stories and conversations from across our nation.

Our reach is huge.

I think it's like 12 million people a month or maybe 12 million people a year, don't quote me.

But our goal is for all Australians to interact with us in their lifetime.

So our job in product design and engineering, it's really important that we make products that are not only just desirable and engaging, but we also think ethical.

And this brings us to Autoplay.

Not all features are created equal.

Some features are revolutionary.

And I think simple old autoplay is one such feature.

Yep, that's right, autoplay.

I think it's revolutionary.

We think it's revolutionary.

Do you remember a time before autoplay?

Let me take you back.

I remember turning on Netflix and there'd be like a thousand million genres like Moody, Monday, Blues.

And I would choose a movie or a show.

I'd watch it.

It would finish and I'd get taken back to the home screen, and then I would choose what to do next.

It was a very active relationship with Netflix.

I was in the driver's seat.

Do I wanna watch another show?

Do I wanna watch another movie?

Do I wanna turn the TV off?

But Netflix being, the superpower, it is, was keen for me, for you, for all of us to keep watching.

They built KPIs around it.

They made the what to do next decision, really easy.

They saw binge behavior and they tapped right into it.

Netflix introduced autoplay.

Now autoplay as a concept is not new, but autoplay in this context was.

By introducing autoplay into the video streaming environment, I believe it changed many things.

It was a bit of a butterfly flaps its wings moment.

Autoplay changed mine, yours, everybody who's watching's relationship, from being an active relationship to being a passive relationship.

With the footprint that Netflix has, it actually changed worldwide expectations of video streaming environments overnight.

I am now not content with a weekly episode of a show that I like, which is how I watched Sopranos for eight years.

Every Thursday I waited for Tony.

But now when I'm choosing content, I want there to be a series of series available.

Not just one.

Designers designed it, engineers built it, and we shipped it.

And everything has changed.

Up to 30 percent of a child's waking time is now spent in front of a screen.

I'm just going to let that sink in a little bit.

That's one third of their daily lives now spent in front of a device.

And studies have shown us that increases in screen time can lead to increases in poor sleep, learning, development and behavioural problems.

Now, at the ABC, we lean into making audiences experiences exceptional.

And we saw autoplay being adopted by our peers.

We jumped on.

We designed it.

We built it.

And we shipped it.

Now, we're not here saying that autoplay is a dark pattern.

That's not what we're saying.

It's very useful when finding what to watch next.

But at the same time, in the kid's space, we saw a rapid adoption of on demand content and a shift to TVs as the main device for viewing.

So as a team, we wanted to understand kids and parents behaviours, needs, goals, and even pain that had come along with that shift.

And what became clear from this research that was championed by our kids team, in Annie Kronis, Annie Jackson, and Soph Brown, was some of the pain that autoplay had accidentally introduced.

Particularly when parents are trying to limit screen time.

And when they're trying to encourage a variety of viewing.

So here is an all too common story that we heard when we were doing the research.

And I'm sure every parent in the room can probably relate to this.

Jack and her dad have come home from school.

It's late.

He's got to get the dinner on.

So they agreed to three episodes and chuck iView on the TV.

That gives dad time to prep.

Fantastic.

But he's in the kitchen and he's constantly having to listen out for the third episode ending.

Dinner's ready.

On the table.

Fantastic.

He realizes he's missed the end credits.

The fourth episode started, he goes to turn the TV off, and all hell breaks loose.

There's tears, there's tantrums, dinner's ruined, and going cold on the table.

We know how hard it is for children to self regulate, so how do we encourage parents to have better conversations with their children about screen time?

So the team took this way, away, and the concept that resonated with us wasn't a new one, but it was a concept called playlists.

We took this audience and it resonated with them as well.

Now playlists solve a primary pain point.

That loss of control when you don't catch it at the right time by actually stopping.

It also stops that dreaded loop of the same content, or the Peppa Pig hole I showed you before.

By providing a variety of content, it actually helps us as the ABC showcase our amazing stage based content.

That empowers our content makers and our editorial teams.

By providing clear, upfront links of time, parents can have better conversations and make verbal agreements about when the TV should go off.

So through this design reframe, the experience has changed from one of high stress for Jack and her parents to one of high value.

And what I wanted to illustrate with this case study is that by adopting features from our industry, can seem useful and even practical.

It's up to us to make sure that we don't introduce unwanted side effects.

It's up to us to think creatively about the problem space to iterate and improve the experience for our users.

Oh, and by far the best feature of this playlist is that when it stops, Croc says goodbye.

It's a much easier conversation for all involved.

See you later, alligator.

So when we think about autoplay, we think about it like this.

A decision was made, a pattern was created, and now this new pattern has changed content consumption forever.

Binge watching a series, a behavior completely boosted by autoplay, has changed the way we consume content.

It has changed the way that content is commissioned.

It has changed the way that content is produced.

And I don't think that anyone realized that this would happen when they designed it.

And now AI is changing the way that people find or expect to find content.

Content used to be found in newspapers, TVs, radios, but today content is mostly found on YouTube with 2.68 billion users.

We're TikTok with 1.53 billion users and Netflix with 230 million subscribers.

Now all of these players use AI in determining which content to show you.

They all share your data and they all want you to stay in their ecosystem until the very end of time.

All of them are changing your, my, everyone's expectations to how, to when it comes to content and what you expect to see, what I expect to see when I land somewhere.

We live in these highly personalized content worlds, and AI has completely disrupted the notion of IA and the content we expect to see.

The better, the more personalized the experience, the more reduced the IA, to the point that like TikTok nearly strips it away completely.

Personalization creates these filter bubbles, of content made for me because it's the content I most likely want to consume.

What gives me pause at night is how different would my world be if I was only served the content that was determined by my 12 year old self, then my, what would my Netflix, what would my YouTube, what would my news homepage look like?

What would Phil's look like?

Would my interests still be the same?

I don't think that the people designing these, cleaner screens mean to or meant to reduce our access to information.

I believe they're making seamlessly engaging experiences.

I think they're driven by their business KPIs.

It also makes me think about filters, right?

The very first dog eared filter.

I don't think the people who designed that actually knew or thought about the implications that filters are now having, particularly on teenage girls.

But all of these things were designed, all of these things were developed, all these things were pushed out into the world.

And the unexpected consequences are what keeps us up at night, maybe especially as parents.

I ask, are we designing a good world?

Are we designing with enough consequence?

I don't think it's all doom and gloom, and I think we've heard lots of people play around this space.

I actually think as designers, engineers, product managers, and with our ability to sit in ambiguity and our ability to play, we can actually harness the excellence of AI.

And with well thought through human focused principles, we can mould a future that actually thinks human first.

We are the people with the authority to research, to design, and to build better.

So what does this mean for the ABC?

In digital product we have a very interesting role to play.

While we don't want to be left behind by not leaning into personalization, we have to balance innovation and new tech with the editorial standards that people expect from us.

We're not here to add to your filter bubble, but we also don't want to make it so difficult that you can't find the content that you want.

It's a tricky balancing act.

Through understanding this balance, we have created these principles, which were created by our audio team, created by Danny Webster, who's sitting in the audience, and then have been since adopted and iterated on by our iview and news teams.

We'll have all four principles up in a moment if you wanted to take a photo.

So our first principle is breadth, not bubbles, ensuring that people are presented with the content as it is.

Outside of their bubble.

Editorial priorities need a human touch.

Leveraging the skills of our incredible content makers and making sure there is space for them in the news that matters most.

Allow our users to feel in control of the content they receive.

Always give our audience agency.

And data fed, people led.

It's definitely my favorite, I think, because it rhymes.

We will always make the decisions.

We won't be fed by the algorithm.

We'll definitely be informed by algorithms.

But we won't be led by them.

These personalization principles allow us to remain creative.

They allow us to create personalized experiences, but they all ensure that we are thinking about the human impact and how data will be used and when it will be used.

So how do these principles land in practice?

So I'm here to show you two case studies, some experiments that we've been running in iView.

One from the kids space again, and the other is a more adults experience.

So we know from research that children watch based on their developmental stage.

And in fact, our curators have known this for a very long time.

They go out to purchase content that meets these stages.

So young children, or puggles as we call them at the ABC, preschool, lower primary and upper primary.

The challenge for us as a product team is how do we know that we're getting the right content in front of the right audience.

So this is where the recommendation engine and its team up with our content curators and iview to help solve this problem.

Curators bring in the human touch for this experiment.

We asked our content curators to go through and tag up all the content into these four stage buckets.

Then based on a child's viewing history.

And using the recommendations engine, we can determine which bucket is appropriate for which child.

Some of the results of this work?

The results are we're actually able to create a much more one to one experience that meet the individual child's needs based on their stage.

We're also able to create breadth, not bubbles in this example.

If you've ever asked a four year old how old they are, they will very quickly tell you that they are four years, five months, three days old.

Children always aspire to be older than they actually are.

And this aspiration plays out in the content that they choose as well.

So what we did was we created an additional bucket called shoulder buckets.

And these live between each of the stages.

So using a child's stage, We can start to introduce and recommend shoulder content that's appropriate for the stage above it.

And as the child starts to watch more of this content, we can grow with them.

Now, our initial experiments in this space have shown really strong results, and we're hoping we can scale this out into the future.

Now, for a more adult example.

It might surprise you to know that the very first feature area that you turn on iView and you get delivered back to you on your TV is a one to many experience.

That means you can turn on and see news when you're actually looking for a drama, You can see Bluey when you don't have children at home.

So what we're doing is we're experimenting with personalizing this space based on your viewing history and your affinity to particular content types.

This here could be MC's experience as a music fan.

Mine would be different again, and yours would be unique to your interests.

So again, we're working with the curation teams.

And the recommendations teams to help solve this problem, defining different rules for the variants that we want to test and learn with.

So humans doing the defining up front, and the machines just finding the right content for the right audiences, and not the other way around.

One thing we are very mindful of, is important and breaking news events.

So early on MC touched on the history of the ABC, and we know that Australians trust and turn to the ABC in times of critical need.

So with this in mind, we actually need to consider much broader scenarios than just any other entertainment platform.

What should happen to these recommendations during a bushfire, an election, a flood?

During these times, audiences just need a way to get to critical information and fast.

Now, we don't have all the answers here, but with this experiment in mind, we've built in controls that switch out recommendation content for curated.

So our human curators can get people to the streams and shows that they need in critical times.

AI and its application in media are endless.

Thrilling and totally intoxicating.

Being able to make content multilingual is happening already.

We are already seeing a whole new era of movie production, where green screens are replaced with virtual sets.

I'm really excited about what our jobs could look like in five years, in two years, tomorrow.

As a team and as a company, we've started to play with all of these new tools as they come to market.

We stretch them, we debate them, we play with them.

As people making things.

I think it's really important that we don't like shy away from this tech, which was my kind of initial thinking around kind of generative AI.

I was like, we have being excellent machine learning team people who can like do all of that thinking.

But it, as I've come to grips with it, actually it's not up to them.

It's not up to one team or one person.

I need to understand it.

We all need to understand this tech.

And we need to play with it, even when it's really uncomfortable.

So as we said at the beginning of this chat, we believe that our role in representing the human actually makes us some of the most relevant people in the room to help with decision making.

When we bought Autoplay in, maybe we did it so quickly we forgot about this.

So when we're starting on the journey with AI and personalization, we're thinking carefully because we know how important it is to get that right.

In a world where the likes of Geoffrey Hinton, the godfather of AI, is asking us to slow down, I'd ask the audience here today to do four things.

Let's adapt over adopt features.

We can all do better than just copy.

Have principles to guide you, so you don't let the tail wag the dog.

Make time for critical thinking and slow your decision making down.

And finally to question, what will the impact of what I'm working on now, be on the future?

And as the product people in the room, let's creatively and consciously design a better world together.

And, for anyone else that's in the room who's interested in how we're using machine learning, large language models, and AI, to help not just the ABC, but benefit our audience, please stick around for Anna Dixon's amazing talk is up next.

Thank you very much.