Why hasn't the 'kitchen of the future' arrived yet?

For years, we’ve been sold on the promise of the “smart kitchen” and the idea that technology would fundamentally transform the way we cook. Rosie the Robot and a smart home took care of all of the Jetsons’ eating needs in that futuristic, 1960s-era cartoon. In the tech-topian film 1999 A.D., also from the 60s, an impeccable mother presses a few buttons in her robotic kitchen and a computer quickly figures out the best meal for her family’s cravings, caloric restrictions and nutritional needs, and then whips it up for her.

It’s 2015. Robochefs should be ordering our groceries and cooking our meals. But the most futuristic thing in most kitchens is a microwave and that’s been around for decades.

The smart-kitchen market is projected to be worth as much as $10 billion by the early 2020s, but right now, the kitchen seems pretty dumb, and unlikely to change anytime soon.

“Many have tried, yet none have ever quite delivered the recipe for success…[It’s the] cultural meme that never seems to die,” wrote Dag Spicer, curator of the Computer History Museum in Mountain View, California. For the smart kitchen to ascend from tech meme to tech staple, the people trying to smarten up the kitchen will have to overcome not just technological barriers, but social ones.

Food and eating are about more than fulfilling our biological need for nutrients—which is why Soylent hasn’t taken over the world yet. Food is community. It’s the connective tissue that binds us to our past and enlivens our present.

While food is a kind of time travel, it turns out that we’re not dedicating much time to making it. Last year, Americans spent an average of just 33 minutes a day preparing food and tidying up after cooking, according the U.S. Bureau of Labor Statistics. Instead of making home-cooked meals, we are taking out and ordering in. Nowadays, less than 60% of meals eaten at home are actually prepared there. The true technological food innovation of our time is the to-go box.

In the last two years, the Internet-connected kitchen gadgets being hawked to consumers have included sushi-making robots, smart coffeemakers, smart refrigerators, smart ovens, smart tea infusers, smart frying pans, smart utensils, and smart food thermometers, to name a few. But maybe Americans are just not spending enough time in the kitchen to actually want to pay the often exorbitant price tag on these tools.

Today, the most innovative kitchens are found not in American homes, but in large hotel chains and cruise ships, says Chris Young of ChefSteps, a food technology company. To help prepare consistently high-quality meals for thousands of people, these operations turn to combi ovens, he says. These specialty ovens have built-in sensors that measure temperature, humidity and oxygen levels. If they detect that your meat is getting dry or overcooked, the software the oven is running adjusts course to make sure you end up with perfectly-cooked food. Unfortunately, the more high-tech ones can cost upwards of $20,000 and require specialized power supplies and plumbing to work.

Young’s home kitchen boasts an Internet-connected sous vide, which cooks food in plastic bags submerged in hot water, but he says his set-up doesn’t compare to that of Nathan Myhrvold, co-author of Modernist Cuisine, the bible for high-tech cooks. Myhrvold has combi ovens, centrifuges and computer-controlled steam kettles, he told me. Ideally, you’d have all these devices talking to each other, scheming together to cook your meal, but that’s not happening yet.

“You want a device that’s not even talking to a human, but that’s still science fiction,” Young said. “Nobody is doing that yet.”

It’s still early days for the smart kitchen. People are trying to figure out how to create a cohesive, intuitive user experience that mirrors what consumers have come to expect from their other tech gadgets.

“The iPod had this wonderfully simple interface that allowed people to browse music in a way they’d never done before,” said Matt Van Horn, the CEO of June, a startup manufacturing smart ovens. “It created a magical experience.”

The kitchen, he says, is starting to see the beginnings of that, as artificial intelligence creeps into the kitchen. His own device, for example, uses computer vision to detect what you place inside the oven, and then suggests how to cook it. It’s a primitive version of a chef’s helper. But even that is still a long way off from Rosie the Robot or the kitchen in 1999 A.D.

To be fair, the kitchen is more of challenge than the iPod or iPhone, which unleashed the app economy, because the devices needed to transform the kitchen cost a lot of money to develop, unlike the apps that make our phones so addictive. So, the rate of innovation is slower than what people have come to expect, and what’s available is very expensive.

Smart ovens that adjust cooking temperatures and times based on portion size and the type of food you’re cooking, like the June, will run you at least $1,000. Pantelligent, a new smart pan, costs $200. Teforia, a high-tech tea infuser, is a whopping $1,300. Maybe your food and drinks will taste better, but at those prices, these gadgets are a hard-sell for anyone on a budget.

The cost is high because building a device from scratch is super costly. Also, the data used to power a device’s “smart” features is expensive. You can’t program a tea infuser to brew the perfect cup of tea or an oven to cook steak just right without consulting experts and doing experiments. That’s pricey because unlike the data that Google and Facebook use to train their AIs to catalog your photos, understand what you say, and predict content you might like, much of it isn’t readily available on the internet for free. Companies need to partner with celebrity chefs and hire scientists and machine learning experts.

And because the data is limited, so are the capabilities baked into smart cooking gadgets. For instance, if I wanted the June, a smart countertop oven, to help me cook pastel Azteca—the Mexican version of lasagna—so that the chicken didn’t dry up and the tortilla layers absorbed the salsa well enough that they were tasty but not soggy, it wouldn’t be of much help to me. Its computer vision algorithms wouldn’t recognize what I was placing on its rack. The reason is that it’s never seen one before. Its algorithms have only been trained to identify 15 different kinds of foods, like bagels, cookie batter, asparagus, chicken, beef, salmon and pork. Plus, the thing works best with preset recipes, like an adult version of an Easy Bake Oven. Stray from this cookie-cutter menu, and you won’t benefit from the handy suggested cooking times and temperatures that are its selling point. So much for my $1,500 investment.

For the smart consumer kitchen to really take off, it needs to operate like our cheap smartphones. Our appliances, utensils and cookingware need to make personalized suggestions, give us instructions on how to cook the ingredients we have, and anticipate our wants and needs—just like Google Now or Google Maps. The smart kitchen needs to provide a magical iPod-like experience that works right out of the box.

If it doesn’t, we’ll get frustrated and keep eating out. To fix that, there’s now a Smart Kitchen Summit, a meeting where technologists, food-startup CEOs and foodies get together to discuss the future of food and what needs to happen for the smart kitchen to become a reality. This year was the first time the event was held. It’s too early to tell what effect it’ll have on the industry. The fact that this was the first event of its kind, says Van Horn, is testament of how embryonic the industry still is.

There’s also the privacy/security issue. Although the smart kitchen dream has been around for decades, many of the companies going into this space aren’t internet companies that know how to deal with cyberthreats. Take a look at the hospitals and medical devices, cars, or smart homes pimped out with internet-connected light bulbs or door locks. It’s not very reassuring. We don’t want snoops to gain yet another set of eyes into our homes through our kitchens, or use our smart cooking tools as pawns in a botnet. Last year, that may have happened. A security company said researchers had uncovered a botnet in which 10,000 internet-connected devices, including refrigerators, were used to send spam email (though some were skeptical, referring to the faulty methodology of the study.)

That being said, there are some signs that the AI-powered kitchen is approaching. IBM has been hawking its Watson supercomputer as a kitchen companion. In June, Big Blue released an app in collaboration with food magazine Bon Appetite that helps people devise new recipes. So far, it’s “thought up” recipes for meals like Peruvian potato poutine, Caribbean snapper fish and chips, and (something less appetizing) an Australian chocolate burrito. As I wrote last month, “Watson is trained on thousands of recipes from Bon Appetite, plus data on food chemistry and human taste preferences. From this data, it develops an understanding of how different ingredients work together to create grub that people will actually enjoy, and so it’s able to give you suggestions of things that are actually edible.”

Earlier this year, Moley Robotics unveiled a robot that learned to make crab bisque on its own, after watching a master chef do it.

YouTube might very well be one of the smart kitchen’s biggest allies. Search for “cooking” or “baking” there, and you’ll get millions of hits. Cooking how-to videos tend to have a common structure, from which computers can pick out common threads and patterns to learn from. Several research groups taught humanoid robots to cook by having them watch YouTube videos and read Wikihow articles.

Those hundreds of thousands of hours of seemingly useless smartphone video of people cooking might actually make the Hollywood vision of smart kitchens come true.

Daniela Hernandez is a senior writer at Fusion. She likes science, robots, pugs, and coffee.