Discover more from FirstNerve
AI on My Mind
“I’m sorry Ave, I’m afraid I can’t do that”
Throw some AI razzle dazzle into your scientific paper and it’s a shoe-in at prestigious journals like Nature or Scientific Reports. Use AI in your business plan and VCs will shower you with funding. And now AI appears to be the new sizzle for flavor & fragrance startups. Consider a few examples and note how the invocation of AI does away with the need for logic or common sense.
Bloviating with Lily AI
Lily AI began as “a product attributes platform that ensures fashion companies and consumers speak the same language.”
Initially, the female-led company looked to build an emotional intelligence-powered shopping experience, available through an iOS app, that used machine learning to understand a woman’s preference and body perception to offer confidence-boosting recommendations.
Sort of a virtual shopping buddy! Well, a round of Series A funding also
allowed the company to pivot from its earlier app-based solution toward an enterprise product offering that focuses on creating customer-centered product taxonomies to improve e-commerce site search, demand forecasting, merchandise planning, product recommendations and discovery as well as search engine optimization and marketing.
Apparently that Series A money allowed Lily AI to buy a shit-ton of buzz words. And to close a $25 million Series B round. Here’s the current mission:
“The language of the consumer is basically where we use deep image recognition technology [and] industry-leading AI capabilities [to] read images and text, extracting attributes that matter to the end consumer,” Gupta, the company’s chief executive officer, explained to WWD.
Sounds awesome, if a tad circular.
Finally, a Chilean unicorn
Today, Chilean food tech company NotCo is selling plant-based milk made with pineapple and cabbage juices, plant-based burgers that contain bamboo fibers, and a new chicken product formulated with corn flour and strawberry extract.
Bamboo fibers—scrummy! And I can never get enough cabbage juice. But co-founder and chief technology officer Karim Pichara is just getting started. In the future,
NotCo could be selling food that tastes like falling in love or the carefree experience of childhood. Pichara, a creator of the company’s powerful Giuseppe AI technology that finds ingredient and function matches between traditional animal-derived products and items in the plant kingdom, said new patents are bringing new capabilities to the system.
Behold the awesome power of Giuseppe AI as it matches milk with cabbage juice!
Perfume from an ugly monkey
NFTs are mostly images of ugly monkeys, i.e., Cabbage Patch Kids for crypto-bros. The new hotness is to link an NFT to a fragrance subscription using a soupçon of AI.
The fragrance is Cyber EDP and it’s billed as
the first unisex fragrance with an illuminated label, embedded printed electronics and a focus on sustainable materials. Inspired by scif-fi movies and AI, the scent has energizing headnotes and a heart of incense with notes of zen wood and amber.
What makes the Cyber EDP NFTs so hot? They
are redeemable with a physical edition of the fragrance which incorporates a printed-electronics label that blinks in red, with 10 limited editions available. First edition owners of the NFT will receive the fragrance signed by the creators and additional hi-resolution versions of the artwork.
Hey kidz, collect ‘em all!
How AI will replace the nose. Or not.
Techie hubris assumes that the right combination of chemometric data and machine learning algorithms will let us predict what a molecule (or a mixture) will smell like. There are a lot of conceptual problems with this approach.
That doesn’t stop the techies who, like their algorithms, just keep grinding away. The latest iteration, by Kushagra Saini and Venkatnarayan Ramanathan, appears in Scientific Reports. Read it for yourself and be amazed at results like this:
The good news is that the tide may be turning. A new paper by Ann-Sophie Barwich and Elisabeth Lloyd in Frontiers of Neuroscience takes skeptical look at whether machine learning can “crack the code in the nose.” The authors argue that the lack of success with computational models won’t be solved with more data or better algorithms—not as long as the models ignore the biological features of the olfactory system (receptors and psychophysics). It’s refreshing to see this common sense contrarian view out in the open. More wet biology and sensory science! Less onanistic number crunching!
Kushagr Saini and Venkatnarayan Ramanathan. (2022). Predicting odor from molecular structure: a multi-label classification approach. Scientific Reports 12(1), 13863.
Ann-Sophie Barwich and Elisabeth A. Lloyd. (2022) More than meets the AI: The possibilities and limits of machine learning in olfaction. Frontiers in Neuroscience 16:981294.