Issue #96 | AI in Food Systems (Beyond the Hype) | Nuclear Science in Food Testing | Event Annoucement | A 🤨 Food Fraud Paper |
2023-07-10
Welcome to The Rotten Apple, an inside view of food integrity for professionals, policy-makers and purveyors. Subscribe for weekly insights, latest news and emerging trends in food safety, food authenticity and sustainable supply chains.
AI in Food Systems - actual examples, without the hype
How isotopes are used in food testing
Event Announcement: July Meetup
News and Resources Roundup (food safety news without the boring bits)
The food fraud academic paper that made me 🤔🤨😐
Food fraud news, incidents and updates
“A paid subscription to Rotten Apple is like hiring another person to give you an extra hand for food fraud mitigation." Deepa, QA Lead.
Thank you, Deepa for your lovely note!
Hello, welcome to Issue 96 of The Rotten Apple, where I cut through the fluff and hype of AI in food systems to bring you actual use cases.
This week I grappled with whether I should share news of an academic paper on food fraud that is so flawed I thought it might be better to ignore it and hope it withers in obscurity. Check the food fraud news section to find out what I did.
As usual, in this issue, there’s a link to food safety news and resources from around the world, including a poo-related recall, aflatoxins in corn and a free meat webinar. Plus food fraud incidents at the end for paying subscribers.
Reminder: Your suggestions are needed. Tell me what you would like more of, less of, and everything in between for this publication by visiting my suggestion box. Open for a limited time.
Karen
P.S. Massive 👏👏👏 shout-out 👏👏👏 to Helen from Australia who is our latest top-tier, extra-special Good Apple supporter. Helen, you rule! Helen’s subscription supports scholarships (free subs) for students, yay.
This newsletter would not be possible without paying subscribers like Helen. Learn more about paid subscriptions and sign up here.
Cover image: Trending Keywords in Food Fraud Research (source)
AI in Food Systems
The Good, the Bad and the Frustratingly-vague
AI is all over the mainstream food media like a rash at the moment, and so many AI-in-food articles are thinly disguised public relations (PR) exercises containing vague platitudes with no juicy details. It’s frustrating.
Reminder: AI stands for artificial intelligence, computer-generated algorithms and models which enable machines to learn from data, make decisions and solve problems. Machine learning, language processing, computer vision and robotics are all types of AI.
Among the hundreds of AI-in-food-systems stories that have come across my desk in the past three months, there have been just three that described actual use cases in a meaningful way. Here they are.
AI helps develop a food fraud analytical method (the good)
A group of researchers at Queens University Belfast used data fusion processes to discover which analytical parameters were best for differentiating salmon samples. Machine learning, a type of AI is applied to data fusion to improve its performance.
Their paper, published last month, described how computer-learning methods were used to first combine raw data from two spectroscopic analytical methods, REIMS iKnife and ICP-MS, and then choose which parameters were optimal for distinguishing between salmon samples from different origins.
The result was a new method that could successfully identify the geographical source of salmon from Alaska, Norway, Iceland and Scotland as well as whether they were farmed or wild-caught.
A carbonated beverage company designs a product with AI
The development team at a Swiss beverage company used the publicly available AI apps ChatGPT and Midjourney to design a new soft drink.
ChatGPT is a text-based AI tool that can generate plausible-sounding texts including blog posts, marketing copy and product formulations using large language models training on millions of strings of words.
Midjourney is an image-based AI tool that mimics images it finds on the internet and combines them to create new images in response to prompts from the human user of the tool.
The product design team at the soft drink company used the AI tools as follows:
ChatGPT suggested a formula for a low-sugar, vegan soda with ‘health benefits’ which included lime juice, haskap berry juice, ginger juice, chicory root powder and cane sugar;
ChatGPT suggested names for the beverage, including the one that was ultimately chosen by the human designers;
Midjourney to create an illustration for the labels.
These tools allowed the team to develop the product in two days, according to Food Navigator.
Two days huh? The product dev techs who had to source the ingredients and make those ingredients into a stable, economically viable and tasty drink might disagree!
Supply chain innovations
Innovations that make use of AI in food supply chains can improve efficiencies, according to Telefonica.
AI can improve the decisions used for forecasting the demand for food items, which means that food companies can produce the correct amounts when needed.
AI-assisted algorithms are used to optimise food transport routes which reduce fuel costs.
AI-assisted inventory management systems work in real time to streamline warehousing processes.
AI completely transforms all stages of food production (the bad)
According to an article published in Hotelier India,
“The F&B industry has always been at the forefront of incorporating AI to enhance its operations, manage inventory, improve food safety and more… transforming how food is produced… in all stages of the supply chain.”
Our industry at the forefront of AI? AI transforming all stages of food production? Yer right!
This made me a teeny bit suspicious. I asked ZeroGPT, an AI-writing detector, whether it thought the article was written by a human. It said no; there is a 100% likelihood it was AI-written.
I wouldn’t have a problem with AI-generated blog posts if they were factual. But they are not, and this is bad.
🍏 For more on how AI-text-creation tools like ChatGPT invent ‘facts’, check out my experiment and cranky rant in Issue 83 🍏
Cultivated meat company uses AI to optimise for flavour and nutrition (the frustratingly vague)
The first three stories above describe real world use-cases for AI in food systems. That’s just three out of hundreds. Almost every other AI-in-food story published in the last few months is frustratingly vague.
Like this one, in PRNewswire, which was probably written to get the futurists salivating and generate investment funding.
It says that a cultivated meat company, Cult Food Science will use AI for taste and nutrition optimisation. It will do this by working with a partner company that uses a “proprietary approach” in which “machine learning is utilized to design the best taste experiences and determine how those tastes should be implemented in [early stages of product] development.” No more details are provided.
Righto.
Conclusion
To summarise, there are certainly opportunities for professionals in the food industry and analytical sciences to work faster and find better solutions using AI-assisted technologies. But I’d like AI-focused news items to contain more meat and less fluff.
For a link to a free webinar on AI in food, see this week’s food safety news roundup
Nuclear Science in Food Testing
Introduction to isotopes in food analysis
An isotope is a variation of a chemical element. Different isotopes of the same element have different numbers of neutrons in their atoms, but the same number of protons.
For example, all carbon atoms have 6 protons, and most carbon atoms have 6 neutrons. A few carbon atoms have 7 or 8 neutrons. Together, these three types of carbon are the carbon isotopes C12, C13 and C14. Most carbon is carbon 12, which has 6 protons and 6 neutrons (6+6=12). Carbon 13 and carbon 14 occur less commonly in nature.
Isotopes can be unstable or stable. Radioisotopes are unstable isotopes and shed particles and energy in the form of radiation.
For example, uranium 235 is a radioactive isotope (radioisotope) that emits radiation. Other isotopes of uranium, with different numbers of neutrons in their atoms, are more stable and emit less radiation. Carbon 14 is a radioisotope and undergoes radioactive decay. This radioactive decay is used in carbon dating in archaeology.
The ratio of isotopes of various chemicals varies from place to place. These differences can be used in food testing.
For example, hydrogen and oxygen isotopes can be found in all food plants and food animals, which take hydrogen and oxygen atoms from their environment as they grow, in the form of water.
Water in different places has different ratios of hydrogen isotopes. Comparing the hydrogen isotope ratios in an unknown food sample with the ratios in samples from a known location allows for the unknown sample to be ‘matched’ to the known location or flagged as unmatched.
In food fraud testing this could take the form of coffee labelled as ‘grown in Columbia’ being compared to a database of authentic Columbian samples to determine if the isotopic fingerprint is similar or different. If the sample has a different isotopic fingerprint, it probably isn’t from Columbia and could be declared fraudulent with respect to the claim ‘grown in Columbia’.
This three-minute video explains the process in plain English.
Read more about isotopic testing for food fraud detection here:
What Is Food Fraud, and How Can Nuclear Science Detect It? | IAEA
Event Announcement
Earlier this year I launched live events as an experiment for 2023, alternating between live training sessions and informal meetups. The informal meetups haven’t been well-attended and while I enjoyed them, they don’t seem to be benefitting you, my lovely readers. So…
There will not be a meetup this month. The next event is a training session:
Food Safety Culture Training (1 hour)
the session will focus on audit-friendly tools and strategies for compliance with food safety culture requirements
Save the date: August 17th UTC 22:00 (London time) | Click here to convert to your local timezone |
More information will be added to our Live Events Page soon.
Food Safety News and Resources
Click the link below for a carefully handcrafted selection of food safety news and resources from around the globe. It’s been expertly curated (by me! 😎) and is free from filler, fluff and promotional junk.
News and Resources Roundup
What you missed in last week’s email
· 3 Real-World Examples of Food Document Fraud
· How to be a Good Food Safety Auditor
· How Bakers Yeast is Made
· Food Safety News and Resources RoundUp
· Food Fraud News, Emerging Risks and Recent Incidents
Below for paying subscribers: Food fraud news, incident reports, and emerging issues, plus a proper human 🎧 audio version 🎧 (no AI voices!)
📌 Food Fraud News 📌
Food Authenticity Network launches a fab new resource
The food fraud academic paper that made me 🤔🤨😐
Conviction for dangerous diet pills
Horizon scanning and recent incidents
A systematic review and bibliometric analysis of food fraud
A recently published academic paper sought to understand the current state of food fraud and food fraud research by analysing peer-reviewed, English-language papers about food fraud.
I struggled with it (big time)…
Keep reading with a 7-day free trial
Subscribe to The Rotten Apple to keep reading this post and get 7 days of free access to the full post archives.