AI can outsmart human beings in many functions and tasks such as mathematical problem solving or at a general level in even abstract tasks like painting or creative writing. But that alone is no yardstick of AI being actually superior to human intelligence. In my mind the real benchmark would be on decision making on the most important decisions that humans and organizations are likely to make.
Even though AI’s advancing super fast, and folks are saying AGI might be just around the corner, there’s still a major issue: decision-making. Even if AGI does show up soon, able to process info and handle tons of different tasks, it’s still gonna be way behind human intelligence when it comes to making decisions – complex, nuanced ones. This is super clear when you look at how AI struggles to get human sensory perception right, and how much we rely on intuition and feelings to make judgments.
Multimodal learning in AI is currently mostly about bringing together stuff from vision (like images and video) and language (text and speech). This lets AI connect what it “sees” with what it “hears” or “reads,” which helps it get a better handle on the context. People are starting to look at bringing in other senses, like smell, touch, and even how our bodies are doing. But this stuff’s still pretty new. So from the very outset if inputs that go into a decision making is fewer then the output is likely to be less accurate.
Smell is complex! Unlike sight or hearing, which are pretty structured (think pixels or sound waves), smell’s crazy complex. Odors are made up of tons of different chemicals, and everyone smells ’em a bit differently. It’s tough to create big, labeled datasets for smells. We don’t have a standard way to categorize or measure odors like we do with images or audio. This makes it hard to train AI models properly. Smell is super subjective and tied to our experiences and emotions. What smells “good” to you might stink to someone else. This makes it really hard for AI to learn and get things right in general. We’re still working on developing sensors that can accurately and reliably detect and tell apart complex odors. What we’ve got now is nowhere near as good as the human nose.
Humans can detect a huge range of odors, even when they’re really faint. We can also tell the difference between really subtle smells, which is super important for stuff like knowing if food’s gone bad or recognizing people. Smell’s really tied to our emotions and memories. It can trigger strong reactions and affect how we act big time. AI doesn’t have this emotional connection. We don’t smell stuff on its own. We combine it with what we see, taste, and feel to get a really rich experience. This ability to mix info from different senses is a key part of human intelligence. Would you leave one of the most important decisions of your life, about falling in love to an AGI?
When we meet someone, smell plays a part in how we see them at first. A nice smell can make someone seem more attractive and likable, while a bad one can do the opposite. For example, whether you like someone’s natural scent or not can be a big factor in deciding if you wanna go on a second date!
Smell can also give us subtle hints about what’s going on. Like, being able to smell fear or stress in someone might affect how we act around them. As a part time chef I know if my biryani seems about right by the smell of the masala without even tasting it. And without smell and taste not being inputs into a model how do we expect AI robots to take over a micheling star restaurant’s kitchen?
Besides sensory stuff, there are other parts of human intelligence that AI just can’t seem to crack. Decision-making, for instance, is often influenced by how we feel and what our gut tells us. We often rely on intuition, those “gut feelings” that nudge us in a certain direction. Warren Buffet has often said that he bets on people. Surely if gut feel has been a factor in decisions of one of the worlds biggest investors, there has to be a critical element that seems missing with analytical and raw data based intelligence. This intuition comes from our past experiences, our brain picking up on patterns without us knowing, and our ability to just get a feel for the situation. Our emotions play a big part in how we make choices. Things like trust, empathy, and how we connect with people can really sway us, especially when it comes to dealing with others or making decisions in business. For example, a lot of business deals happen because the people involved feel good about each other, not just because the numbers add up! It’s tough to teach AI to do this kind of intuitive, emotional thinking. AI tends to stick to logic and hard data.
To sum it up: AI’s getting better at using different kinds of information, but it’s still got a long way to go when it comes to things like smell, taste, feelings, intuition and gut feel. The fact that all these varied sets of sensory and sometimes super sensory inputs go into a complex mix of interactions that lead to the way we make decisions means that AGI in whatever way it manifests will lag behind human intelligence when it comes to the most important set of decisions that we make.
