• Subscribe

Atypical AI food

Over the last decade, AI has infiltrated almost every sector of American industry and most American lives. As of 2019, most US citizens were already using AI in the form of navigation apps and streaming services, like Spotify. That trend is only set to increase, as businesses and countries lay out plans to integrate AI into their development strategies.<strong>Average AI applications</strong> like navigations apps and music players are ubiquitous within our lives now. However, this technology is anything but average, and AI can do a lot more than you might think.

Inevitably, some AI applications and trends become buried beneath media outpourings on economically driven AI innovations in the standard sectors of manufacturing, health, advertising, ecommerce, and surveillance.  

Such has been the case recently, given the media’s captivation with COVID-19-inspired AI innovations in medicine and telework. So, we’ve taken the time to unbury some trends that are not as likely to appear in search results.

For this edition, we've focused on AI innovations within the world of food. 

AI for animal-free animal products

In 2013, Mosa Meats introduced the public to a revolutionary new type of meat: lab-grown, cell-based hamburgers. Over the course of three months, three lab technicians cultured real animal cells in a nutrient medium, producing strings of muscle cells, eventually reaching the large sum of 20,000 strings—enough for one burger.

Producing the world’s first lab-grown burger cost around $300,000. The company, which wants to bring the price down to $11 per burger, recently began seeking machine learning specialists to accelerate its research.<strong>A $300,000 burger</strong> obviously isn't feasible. However, this technology will get cheaper as we explore it further.

A number of additional startups have given life to the burgeoning cultured-meats industry. In London, 2020’s startup Hoxton Farms is leveraging computational biology and machine learning to produce cell-cultured animal fat. Animal cells are grown in a bioreactor and harvested at maturity—a process which they iteratively simulate and improve using a “digital twin.”

The innovation is an important step toward engineering complex meats, like steak, in which fat affects both structure and taste.

AI for plant-based animal alternatives

With cell-cultured meats threatening established production vendors, legislation (like Montana’s “Real Meat Act”), which would limit how the meat is labeled at market, has spread through the states. Plant-based animal alternatives, on the other hand, are not new to the ire of traditional vendors.

But with an ever-growing market for plant-based products, new startups are not deterred, nor are AI-based researchers immune to the draw.

In November 2020, NotCo made its US debut with its AI-formulated NotMilk, which contains such strange ingredients as pineapple juice and cabbage juice. The company’s AI technology, Giuseppe, has since earned the US’s first patent for a plant-based-food AI technology.<strong>Milk from cabbage juice?</strong> AI really is an incredible technology!

The AI creates original recipes for a variety of plant-based animal products, including milk, ice cream, and mayonnaise. After training on thousands of datasets of extant recipes and food components and their chemical makeups, it predicts which ingredients, proportions, and cooking processes would deliver the taste, texture, and colors of the original.

And the results are often surprising. Hampton Creek (renamed Just, Inc.), using a similar technique, found a scrambled egg substitute—mung beans. And NotCo’s chocolate recipe prototype consists of broccoli, goji berries, and mushrooms, among other ingredients.

AI toward cultured milk

Today’s relatively convincing plant-based milk products still leave about 33% of customers returning to dairy because of compromises in taste, creating opportunity for companies like NotCo, Hampton Creek, and the novel Imagindairy, which draws on cell proteins, rather than plants.

Imagindairy, a startup that arose out of a Tel Aviv University professor’s lab, is examining the inherent milk recipes encoded in the genes of cows. These genes can be translated and inserted into other organisms, like yeast, which would then become factories for milk proteins. <strong>Milk</strong> has a very unique taste and mouth feel. Replicating what milk has to offer is no small task.

But translating the language of cow proteins into the language of yeast is not easy. For that, the researchers are drawing on machine learning and computational modeling of molecular evolution. The professor of the lab, Tamir Tuller, has successfully used this protein coding technique in the past with vaccines and medicine.

The technique in question allows for customization; through it, they could direct the genes to continue coding for milk that’s identical in flavor, nutritional value, aroma, and whatever other characteristics they care to preserve, while writing out the expression of cholesterol, lactose, and somatic cells—creating a healthier alternative to traditional milk, without the flavor tradeoff.

However, the researchers assure us it’s not an alternative; it is milk.

AI and the art of fish grading

In fish as in milk, quality and flavor are key.

But while some consumers have spent as little as minutes shopping at their nearest grocers, others have traveled hours to reach Tsujiki Market, once the world’s most recognized fish market, to maneuver through its warren of 900 shops and stalls, alongside thousands of others, in search of the perfect tuna. 

Until the market’s relocation to Toyosu in 2018, the stalls were manned by tuna traders who had spent years mastering the art of tuna inspection, a tradition passed on through generations—a reflection of tuna’s status in Japanese cuisine and culture.

Now an app named Tuna Scope, developed by Japanese firm Dentsu Inc., offers to do the same. The deep-learning AI was trained on over 4,000 images of tuna tails, an area inspected by traditional craftsmen. Slight nuances in the color and sheen of the tail’s flesh offer key insights into the flavor, taste, and life quality of the tuna.<strong>All tunas</strong> are not created equal. Various life factors can increase or decrease a tuna's desirability, and it takes a keen eye to grade these creatures.

Adopted by and tested at Yaizu factory, the app-based technology performed in 85% synchronization with experts with 35 years of experience. Dentsu reports that the best tuna from the trial was sold at a Tokyo Station restaurant as sushi, where approximately 80% of customers reported it was better than usual. 

While the app comes at a prescient time, when many buyers are unable to visit wholesalers in person because of COVID-19, buyers for some higher-end restaurants say that they will always prefer to do it by eye.

With innovation comes change, and many have ethical concerns, especially in the way that cultured animal products could upend the farming industry and the lives of its workers. Keeping these needs in mind will be vital as we move into this next chapter in AI technology.

Read the second article in this series. 

Join the conversation

Do you have story ideas or something to contribute? Let us know!

Copyright © 2023 Science Node ™  |  Privacy Notice  |  Sitemap

Disclaimer: While Science Node ™ does its best to provide complete and up-to-date information, it does not warrant that the information is error-free and disclaims all liability with respect to results from the use of the information.


We encourage you to republish this article online and in print, it’s free under our creative commons attribution license, but please follow some simple guidelines:
  1. You have to credit our authors.
  2. You have to credit ScienceNode.org — where possible include our logo with a link back to the original article.
  3. You can simply run the first few lines of the article and then add: “Read the full article on ScienceNode.org” containing a link back to the original article.
  4. The easiest way to get the article on your site is to embed the code below.