Close

Idealog eats robo-food

No stranger to questionable culinary choices, digital editor Ben Mack tries a dessert designed by IBM's Watson in partnership with Fonterra - and learns a little bit about what a future full of "robo-food" might be like.

Let’s just get this out of the way: the idea of eating pavlova with cheese on top is pretty weird to approximately 99.99 percent of the population. Even weirder: if that “cheddarlova” is designed by an artificial intelligence.

And weirder still: if that artificial intelligence just so happens to be IBM’s Watson. You know, Watson? The Jeopardy-playing computer that inexplicably thought the US city whose largest airport was named after a World War II hero was Toronto (it’s Chicago)?

And that’s not even the end of the weirdness: all this happened with the support of dairy giants Fonterra that resulted in enough cheese/pavlova strangeness for several hundred people to indulge in the oddity for themselves.

Oh, and here’s the weirdest thing of all: it tasted bloody good.

This all requires some explanation. You see, to celebrate the opening of its new offices in Auckland’s Wynyard Quarter (a pretty snazzy complex Idealog got to check out last week), IBM hosted a series of events to help the community get used to the fact one of the world’s largest companies was setting up shop in an area that already hosts some pretty big businesses (like Microsoft, HP, Datacom, Air New Zealand, Fonterra and ASB). And the culmination of all that: a free feed for whoever wanted to pop by and eat food whose recipe was created by a computer programme.

I’ll admit I was a bit sceptical. After all, when I went spent several days aboard a cargo ship these past Christmas holidays, I was concerned that no-one was actually steering the 360-metre vessel while it was at sea – it was all done by a sophisticated computer programme. Food-wise, I was also worried at a recent degustation in the dark that I’d accidentally ingest vegetables (I hate veggies).

Hearing how Watson came up with the recipe also was not the most promising. According to IBM New Zealand analytics and cognitive principal Isuru Fernando, Watson was fed more than 10,000 recipes from US magazine Bon Appétit, as well as heaps of digital material about Wynyard Quarter and Aotearoa. “We thought it’d be interesting to create a dish that really resonated with the Wynyard precinct,” he explained. “It’s a true collaboration between man and machine. Working outside the confines of human taste and experience, the AI creates recipes that challenge our typical ideas of what will taste good.”

One of those sources for information that Watson relied on: Wikipedia.

Uh-huh. Wikipedia. You know, the website that, if you ever dare to admit to a university professor that you used it for research for an essay, you probably will be failed automatically for being foolish enough to believe anything you read online? The same website that formerly, in its entry for George W Bush, declared in the first paragraph that he was a space alien? Yeah, that website.

So, needless to say, I was concerned. Isuru assured me I shouldn’t be. After all, he said, this was about a whole lot more than the “future of food” – it was about how AI could one day run our lives.

That was even more concerning.

“AI is going to be permeated throughout humans,” he told me. “Everything we do will have some sort of cognitive computing.”

Streaky clouds whizzing by overhead at speeds faster than almost any I’d ever seen in clouds, that was genuinely a reason to be frightened for the future, I thought. I’d never seen Terminator, but I knew the premise of humans relying too much on AI, and the AI deciding it didn’t need to rely on humans. It’s also an anxiety a lot of people undoubtedly felt when another IBM computer, Deep Blue, defeated chess grandmaster Gary Kasparov in 1996.

But this was different, Isuru claimed. It was all about learning how computing could help personalize our lives, through things such as healthcare tailored specifically for us, with information shared between doctors and caregivers via the cloud. Having Watson create a recipe could also help people understand how AI can “serve” humans, he said.

It’s also a view shared by Fonterra’s heard of digital design, Dominic Quin – and one reason he said Fonterra was more than happy to team up with IBM to see what Watson could cook up. “We’d rather be disrupted by ourselves than others,” he told me as I noticed out of the corner of my eye that no-one was retching from Watson’s recipe yet. “It’s about keeping ahead.”

Watson’s personalization capabilities could be a massive benefit to Fonterra, Quin explained. For one, it could help farmers manage their herds more effectively, from knowing what pastures are providing optimal nutrition for their cows to even figuring out ideal milking times and individually managing each cow’s health. And Watson also helps Fonterra get the word out that it’s far more than a co-operative that receives milk from farmers and then sells it. “We’d like to think of ourselves as a technology-driven company.”

But the proof would have to be in the proverbial pudding – or in this case whatever culinary concoction Watson had come up with. Isuru and Dominic both assured me it tasted fine, though it was “something you normally wouldn’t think of.”

Watson did have a little bit of human help, though – at least in the form of the person who took the recipe and physically put it together. But even Keith McDonald, a professional chef and Fonterra’s global category and innovation manager, admitted he was a bit sceptical at the idea of a computer-generated recipe, and had doubts the flavours would work.

“I was truly amazed by the taste of the dish. Working with technology like this has opened my eyes to a whole new level of cooking,” he said. “As chefs we’re always creating new menus, looking for ideas to push the boundaries and find the next best thing in food. You have a group of flavours you know that work. It’s one of those things you have a hard time to envisage how it would taste.”

From left: Dominic Quin, Keith McDonald, and Isuru Fernando.

By now, dear reader, you likely have a single, burning question: just what the bloody heck was this recipe?

It was pavlova. With cheese. Seriously.


Known as “cheddarlova,” the recipe also had some fruit, chili, and lime. But the combination of pavlova and cheese was what shocked most people – McDonald included.

“To do something like this with pavlova is… almost unheard of. We genuinely thought there’d be people who’d spit it out.”

Ironically, I was unfazed. Famed for decidedly bizarre culinary choices – I’d recently put some melted Colby cheese, a fried egg, and surimi inside a cheese scone for dinner and decided it was the tastiest thing I’d had all week – I was more intrigued than anything else. A robot after my own heart. Maybe Watson wasn’t so bad after all.

McDonald said the cheddarlova recipe was selected by IBM and Fonterra out of four that Watson came up with. McDonald said it grew on him, too. “I see how working with AI can help us come up with plenty of challenging and trend setting flavours – like salted caramel and now well, cheddarlova.’

Mouth watering like Pavlov’s dog, it was finally my chance to taste it. Isuru, Dominic and Keith watching intently, I gingerly picked up a piece of slightly spongy pavlova – taking great care not to spill any of the cheese or fruit heaped on top – and plopped in into my mouth. A sugary explosion went off like a bomb. A river of savoury gooiness followed.

It was bloody delicious.

So much for machines not knowing how to cook.



For anyone that wants to try it at home, here's Watson's cheddarlova recipe:

•Sharp cheddar cheese
•Chilled heavy cream
•Shredded coconut
•Peeled mango
•Light brown sugar
•Passion fruit juice
•Chilli powder
•Corn starch
•Egg whites
•Golden raisins
•Bay leaf
•Cream of tarter