Imagine this scenario for me. You finish a sweaty cardio session at your local gym, and after hitting the showers, you swing by the gym’s smoothie bar for the perfect recovery smoothie made just for you. Your unique workout data is synced to your diet plan via the powers of artificial intelligence and machine learning, and it generated the perfect combination of ingredients for your recovery smoothie to maximize your weight loss and muscle recovery goals. This magical intelligence can also consider your sleep, insulin, and stress levels, optimizing your diet at every turn and suggesting the right cocktail of foodstuff to improve your health and quality of life. Does this scenario seem like a sci-fi television episode from when we were kids, along with flying cars and teleporting?
No, this is what Deliveroo’s “Snack to the Future: 2040” Report, released in early July predicts about food consumption by 2040. According to the report, the global personalized nutrition market is valued at £14.5 billion in 2023 and is predicted to grow to £66 billion by 2040. That’s no small feat - there will have to be a significant innovation in personalized nutrition solutions, and the brand has just the term ready to coin for it: “me-ganism” (you heard it from me first).
For some of us, this AI-driven dream is one we’d like to chase. If you have had enough trying to think through menu planning, grocery shopping, and meal preparation, passing the buck to artificial intelligence may feel like a heaving sigh of relief. For others, the mere suggestion that AI infiltrates our minds and takes over our food choices is closely knit to a nightmare we once had. However, we aren’t far off from this becoming a reality.
Today, ChatGPT can be doubled as a chatbot to ask how to create a diet based on certain parameters and break that diet into ingredients to order online. Machine learning aspects are already being used for online grocery shopping, using predictive models of the food you may prefer based on your past choices.
Regarding physical evaluations, we already have technology that reads our insulin levels before and after meals to determine how and what to eat retrospectively. Some companies offer to analyze your feces (if you’re willing to collect it and drop it off at your post office) to determine the health of your microbiome, to provide bespoke diet plans (which, admittedly, are questionably effective at present), and, bordering on the obscure-but-fascinating innovation is the award-winning personal toilet insert called U-Scan that analyses your urine for a variety of metrics to determine your health. The interconnectivity of this technology is limitless and mind-blowing and not as far-reaching as we think. We are a few inquisitive minds away from a breakthrough in this technology, and no matter if it conjures fear or excitement, it’s a unique time to be alive.
However, not to pour water on the fire of your hopes and dreams to never meal plan again, significant gaps exist between what technology exists now and what needs to improve before AI can take over our meal planning.
Little available knowledge
The first thing to consider is that, as a species, we know remarkably little about nutrition and how it can impact, heal, and biohack our bodies. Humans have been eating since time immemorial, so this is a colossal missed opportunity to collect knowledge. Modern doctors in most developed countries receive shockingly little training on food as a preventative medicine and the average consumer relies on the industry’s own marketing tactics to dictate their understanding of what’s good for their bodies.
Influence of the food industry
Therefore, it’s safe to say that we are still learning. To learn, we must research and test. Unfortunately, there is a fundamental flaw in our scientific community where the thrill of discovery is often limited to those who get funding, and it’s not equanimous in its distribution. After all, how many studies do we need on sugar or a new superfood, but rather on how different foods interact to create an individual health prevention strategy? As some studies are backed by some major players with vested interest in the outcome, results may be conflicting when analysing objectively all the existing materials to make an efficient learning platform. The reality is that we need more funding and objective nutrition studies to help make this a reality.
No one-size fits all diet
The third thing to consider is that our bodies don’t respond to diets similarly, and no one-size-fits-all diet exists. The idea of ‘me-ganism” is fascinating, and the glucose monitor is an incredibly interesting way to capture our individuality with food. However, insulin is not the only factor to consider, and our biologies create many complexities that are challenging to navigate. For example, if you suffer from diabetes yet also have a high risk of kidney stones, to what degree can AI ascertain the perfect diet for you, considering a few dietary suggestions may differ within your circumstances?
So far, the world of AI is input based, meaning it’s based on information about decisions that have already occurred. This retrospective analysis can be helpful for some, and they find it interesting to review the data obtained from their devices, and they set out to change their lifestyle habits. Still, when it’s said and done, it’s a reflection of their past decisions, not a real live model of their ongoing decision-making, and this is a highly complex arena to dabble in. Considering this, a budding new AI process called “inverse optimization” has emerged from John Hopkins University. Inverse optimization considers your decisions the “inputs” and determines the objective. For example, if your diet is laden with salt, inverse optimization may look at your current diet and determine ways to lower your salt intake.
But here lies the crux of the issue – machine learning can suggest something to us, but it will mean full abandonment of personal choice. To choose a piece of cheesecake when we want it, even though a salad is suggested on our “me-nu”, or to skip dinner when we know we should have it. AI can only offer us suggestions; it’s up to us whether we adhere to its guidance or not. You can glean all the data you want from your body, but if you don’t have the willpower and motivation to follow through, this information sits as it is: interesting information.
Ultimately, AI-driven diets come from a place of good intention to demystify a complex process to help us live a healthier life, and it will be interesting to see what the actual future of food holds for us. However, I still believe that AI-driven diets and their offshoot technologies are part of the solution, not the cure. No fancy or futuristic piece of technology can force you to eat what it wants you to, and that’s the deliciously powerful part about being human: our right to choose and the luxury to change our minds.
Jen Thomas is a master women's health coach.