What happens if you combine one of the oldest industries to one of the newest technologies? We have seen what can happen when a disruptive innovated changes an industry, for example the transport industry, how many drivers will there be in ten years? One of the next possible targets is the food industry, more specifically; restaurants. This industry has been growing steadily since 2009 and in 2015, food and drink sales in the United States restaurant industry amounted to 745.61 billion U.S. dollars (Statista, 2017). Keeping this in mind it is no surprise that tech giants are trying to settle in this industry. Both IBM and MIT have founded AI-technologies that can revolutionize the cooking industry.
IBM has developed a question answering machine over the past 15 years that is called IBM Watson. Watson got world-wide fame when this ‘supercomputer’ defeated two of the world’s best players in a game of Jeopardy!. It now has further evolved and has multiple applications, one of them is IBM Chef Watson. This software works as follows; in the app you type your favorite ingredients or food that you have left in the fridge and Chef Watson will create a dish based on complementary flavors or recipes. This will allow even people that have moderate tasting ability cook like a chef.
Researchers at the Massachusetts Institute of Technology have created an app that works the other way around. The app lets users identify their food based on a photo. So if you find a photo on Instagram with a nice dish on it, you can put this in the app and it will tell you what ingredients are in there and it will provide a recipe. In testing the app called Pic2Recipe identified the ingredients correctly in 65% of the time. The system uses deep learning to get to know over 1.000.000 recipes and this number only keeps growing. Deep learning is a branch of machine learning based on learning data representations (Bengio et al., 2013). This allows the Pic2Recipe to look at food differently, even in manners unknown to humankind.
Although still in its infancy, food and AI are coming closer together. Next to that, more organizations are getting involved in the food industry and most of them are tech-corporations that are already investing billions of U.S. Dollars in restaurant-industry-specific technologies. So what will this mean for the future? Will there be ‘chefless’ restaurants? Will chefs become more skilled because of the help of Artificial Intelligence in the restaurant industry? Or, will we all be better cooks in future? One thing I do know is that it would be great when fine-dining becomes more approachable.
References
Bengio, Y., Courville, A., & Vincent, P. (2013). Representation learning: A review and new perspectives. IEEE transactions on pattern analysis and machine intelligence, 35(8), 1798-1828.
Brandt, R. (2016). Chef Watson has arrived and is ready to help you cook. Retrieved from: https://www.ibm.com/blogs/watson/2016/01/chef-watson-has-arrived-and-is-ready-to-help-you-cook/
Conner-Simons, A. Gordon, R. (2017). Artificial intelligence suggests recipes based on food photos. Retrieved from: http://news.mit.edu/2017/artificial-intelligence-suggests-recipes-based-on-food-photos-0720
Kleinman, Z. (2017). AI demo picks out recipes from food photos. Retrieved from: http://www.bbc.com/news/technology-40681395
Statista (2017). Restaurant industry food and drink sales in the United States from 1970 to 2016 (in billion U.S. dollars). Retrieved from: https://www.statista.com/statistics/203358/food-and-drinks-sales-of-us-restaurants-since-1970/
Thank you Hidde for sharing this very interesting post. I have recently start looking into new development of the food industry in terms of technology and way of ‘experiencing the food’ (I might write a blog post on that as well). Yet, I always focus of the other side of the market, thus at the customer experience.
Customers in the recent years have been given a large number of tools that allow them to become better cooks. Gordon Ramsey’s books and TV shows, as well as Chef Jamie’s awareness campaigns, have made the knowledge of quality raw ingredients, the will of discover and try new receipts and the need of a more healthy and researched cuisine more common among the mass.
Yet I personally do not think that a robot (or an application) will be able to overtake humans. While cocktail robots are already a thing (http://www.makrshakr.com/) they are so far only able to replicate – in a perfect way, don’t get me wrong – a pre-set recipe. While we saw that AI can overtake these kind of problems, the biggest doubt I have is the fact that the sense taste, smell and touch cannot be simulated by an artificial intelligence yet.
This will therefore, at least in the short term, undermine the ability of these technology to overtake the human experience and ability. This there are no doubt that the implementation of the technology will enhance the average leger of the population in term of awareness and skill.
Hi Hidde. That’s a nice article you have written there. I think this can definitely help to minimize waste. Instead of throwing or letting vegetables go bad, Watson can think of a nice little recipe to use them. I sometimes feel uninspired about what to cook, glad technology can think with me. Thanks, Stephanie
Hi Hid,
Cool article. What I like is that you touch upon many different elements, including food waste. Helping people to combine their leftovers into a nice mail can be of true value.
As food and AI come closer together, you can imagine some cooks being replaced by AI in cheap restaurants for instance. Also, as mentioned some people who do not like cooking or cannot cook properly can use AI to help them with that. For those instances, I see a real market.
However, there are two questions that I have for you.
Personally, I enjoy cooking and like to combine and taste different recipes. Do you think AI will truly ever take over? Many people enjoy cooking and I think that trusting a machine to cook your food all the time is far away. How do you see this?
Then, how often do you think the AI cook has to get it right to be able to be sold to customers? I mean, by machine learning the AI cook can probably get a right meal most of the times, but they can fail miserably as well. In the end, the machine cannot taste anything.
Thanks!