What AI Tells Us About Ourselves

I saw today that Bumble is incorporating into their business model a way for “AI” to “go on dates” instead of people. Selling their customers on the idea that it can go on 600 dates for them, save them time in finding a mate. They even envision the chat-bots being intimate companions to those seeking mates, gathering even more word data around their feelings.

I think, as a business and for their shareholders, this is probably a good move for them, augmenting whatever psychometric profiling they’re already doing.

But a big part of me pauses here.

The data harvesting, it gets ever deeper and more intimate. None of this, of course, is actually AI. It’s still machines collecting data (learning, machine learning) and finding correlation in data. It knows a fair bit about words, which is how humans communicate through computers, but such a small part about how humans actually communicate. If we follow the 55/38/7 rule, 55% of how we communicate is nonverbal, not transmissible across the Internet, 38% of how we communicate is vocal, how we say the words, not ingested by LLM training, and 7% is the words themselves, which LLMs have “learned.”

It knows 7% of how to drive, and everyone wants to give it the car keys. It knows words, and thus can be matchmaker, directly influencing human coupling and the creation of future human generations.

Our call to “AI” to solve our problems, I think it is a clear message of despair and hopelessness from society, a clear call out for help. Everything around us seems to be failing, and this feels like our desperate voice, begging.

I’m reminded of how the Agile community began to react to “scaling agile.” We channel things like Elisabeth Kubler-Ross’s work on grief, from the 1960s, to remind us that change starts with human beings, and we humans respond to change in emotional ways. Jonathan Smart, for example, connects that directly to what happens when we try and affect large-scale process changes in organizations - the larger the change initiative, the harder it is, because of the human response.

Instead, Jon joins other agilist leaders in rallying us around using the network effect to engage human experience at scale to drive change for us - if you’ve ever heard me or other agilists rant about “de-scaling” - this is what we mean.

Any human facing business forms through experiences. Every business I’ve had grew because I, mostly indirectly, fostered positive human experiences. Sure, the business had to deliver utility to be viable, but it was all the intangibles that brought me more business through those successes. As my businesses grew, the humanity within the business, the culture, further fueled these successful customer experiences.

Large Language Models, Large Multimodal Models, none of these things can ingest what can’t be digitized. They can’t help here.

Our digitized representation of reality is not reality.

The map is not the territory.

I urge people to explore machine learning. It’s a powerful tool that can help us in many ways. It will not fix our problems for us, though, in society or in our businesses.

Let’s use “AI” to gather us for more human experiences, not try and have them for us.

Bring it the menial work, so that we and our businesses can blossom together.