What Is Intelligence Without Sense of Smell? by Dr. Timothy Smith

Photo Source: Unsplash

The unbridled expansion of AI and the requisite proliferation of data centers to support and scale the latest models form the basis of the revolution in computing that has transfixed the world. Still, this massive push for artificial intelligence should prompt the question: what constitutes intelligence? Part of the answer lies in perception. The ability to see, taste, feel, hear, and smell shapes our perception of the world and the intelligence we develop to survive and thrive in our environment. Large language models such as Gemini and ChatGPT learned primarily by analyzing the written word pulled from books, documents, and internet content.

Words represent actions, thoughts, things, or experiences, but they are not actions, thoughts, things, or experiences. This distinction matters because human intelligence evolved to help us adapt and survive in Earth's environment, and the sense of smell emerged as the first sense. The earliest creatures in evolution, such as bacteria, developed the ability to distinguish different chemicals in their environment, which helps them find food, avoid toxic substances, and trigger group responses. When a bacterium "smells" a competing colony, it can initiate the creation of a multicellular biofilm to protect itself.

The basic function of smell began with the earliest organisms and evolved in animals and even plants. For humans, the sense of smell connects through the olfactory bulbs in our noses directly to the brain, unlike sight or sound, which pass through. Researchers think this direct connection between smell and the brain contributes to the smell associations that form our earliest memories. It only takes one or two exposures to an odor as a child to remember it for life. Such potent experiences as smell can conjure early memories tied to our first experiences of good and bad smells connected to food, environment, and people. This basic form of learning helps us learn what may be safe or dangerous. Interestingly, of the thousands of smells humans detect, most don't even have a name, which keeps them out of the learning materials LLMs use to develop artificial intelligence.

The miniature camera and microphone so prevalent in our smartphones and computers give computers and robots access to sight and sound, but the olfaction, or smell detection, that occurs at the molecular level in our noses has no miniature equivalent for machines. Chemosensing machines that act like a nose, identifying volatile chemicals in the air, use sophisticated separation and identification techniques such as mass spectrometry (MS) and gas chromatography (GC). MS uses powerful magnets to separate molecules and to determine their identities. GC uses long tubes lined with materials that cause different molecules to travel through the tubes at different rates, depending on their chemical properties and identities.

The GC and MS machines in the laboratory occupy large spaces and require technical expertise to operate. Like many things, companies have worked to miniaturize MSs and GCs to make them portable and small enough to attach to a robot. Currently, the smallest available spectrometer, the BaySpec portable MS, weighs 22lbs and takes up 14.5 x 13 x 9 inches of space, and the even larger Torion T-9 (Tridion-9) GC-MS weighs 32lbs. Given the lightweight and amazing sensitivity of animal noses and insect antennae for chemosensing, robots and computers will not soon have a sense of smell. This fundamental lack of smell in AI divides our types of intelligence and shapes how we react to the world and process our experiences.

Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high-level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.

You can buy his book on Amazon in paperback and in kindle format here.