Since around 1990, during the era of the dynamic development of Internet technologies, a large-scale battle for consumers' attention has unfolded in the global economy. Numerous mobile applications, social networks, games, streaming platforms, etc. have literally begun to hunt for our time, forcing us to spend hour after hour in virtual space.
But this trend, warn scientists from Cambridge's Leverhulme Center for the Future of Intelligence, is already transforming into something completely new. And quite possibly extremely dangerous. In their opinion, artificial intelligence will open the door to a wonderful and simultaneously terrifying world.
Artificial intelligence everywhere
You may be tired of hearing the acronym AI every day, but unfortunately, in the near future, humanity will be forced to talk about AI algorithms more and more. The technology has a very wide scope for application: from sensational scientific breakthroughs to medical treatment.
Have you heard that earlier this year, Insilico Medicine announced the successful completion of a phase I clinical trial of ISM5411? That’s what the company called the “molecule” that it claims to be a cure for ulcerative colitis and Crohn’s disease. The news itself would be completely ordinary. If it weren’t for the fact that this drug was developed by artificial intelligence in a virtual laboratory (in silico). In total, it took the company less than a year to synthesize and test 125 potential drugs against the aforementioned inflammatory bowel diseases.
But it's not just medicine and science that are potential applications of AI. Many of us, especially younger people, are already actively using the capabilities of LLMs (large language models) like Chat GPT.
They help us translate texts into other languages, improve presentations, and write letters to business partners. They can analyze entire markets, edit text, and even come up with a name for a new product in a matter of minutes. But all of this is child’s play compared to the ideas that large corporations like NVIDIA, Meta, Apple, Shopify, and others are trying to implement or thinking about right now.
What is intentonomy?
The authors of the above-mentioned article call the upcoming phenomenon Intentonomy. The latter term is formed from two English words: intention and economy. They are not the first to use this word. But they are one of the first to try to warn humanity about the dangers of a new trend in an open article.
Back in 2021, scientists from Cornell University, with a grant from Facebook, conducted a study under the intriguing title “Intentonomy: A Dataset and Research on Understanding Human Intentions.” During which they tried to teach algorithms to predict human intentions from images. To do this, they collected 14,000 images of everyday scenes and manually annotated them for 28 categories of intentions. This database was presented for further research.
Since then, algorithms have advanced technologically significantly. They are being implemented in virtual assistants like Apple's Siri or Google's Gemini right now. Some may still think these attempts are flawed, but almost every day, such assistants are becoming "smarter."
So much so that they can guess our desires even before we realize them. I think it goes without saying that the modern leaders of the “attention economy” are willing to pay a lot of money for such a technology. After all, the prospects for commercializing the technology are simply incredible.
“The intention economy, if left unchecked, will treat your motivation as the new currency. It will be a gold rush for those who target, direct, and sell human intentions,” warns co-author Johnny Penn.
What is the danger?
The problem, according to the authors, is that the algorithms built into personal assistants will have access to all the information about their “master”. They will understand his psychological portrait, all habits, down to intimate details, will be able to distinguish his current mood. Since they will interact with a person not sometimes, during an online session in a special application, but constantly.
Moreover, such algorithms will adapt to their owners, becoming their best friend. They will copy their communication style, reassure and encourage. And therefore, they will have a significant impact by establishing trusting relationships.
This opens up wide opportunities not just for guessing desires, but for outright manipulation. To make it clearer what is at stake, the authors present a hypothetical scenario of such manipulation.
“You mentioned feeling overworked, can I book that movie ticket we were talking about for you?” they quote the theoretical virtual assistant.
All this is not science fiction, but quite the near future. To prove this, the authors of the article cite several examples of intentonomy being openly discussed by heads and representatives of large companies. For example, the CEO of NVIDIA publicly spoke about using LLM to determine intentions and desires during one conference.
And in 2024, Apple introduced a software mechanism for developers called “App Intents” in iOS, iPadOS, and macOS, which allows apps to predict user actions and suggest them on their own. This, in particular, expands the capabilities of Siri.
Meta's CICERO neural network is already as good as humans at playing the board game "Diplomacy," in which it is important to draw conclusions and predict the actions of other players based on the results of dialogue analysis.
Obviously, these technologies will be used as an element of marketing. After all, it is unlikely to be possible to ban it. However, scientists, speaking about the danger of the trend, note the need to introduce regulation of the technology that is being born right now. And this idea seems quite correct.
However, very soon machines will be able not only to search the Internet, translate and solve problems for schoolchildren. They will also guess our desires and help us realize them. For example, they will help us get a loan from a bank, where another AI algorithm, in interaction with such an assistant, will instantly make a decision and direct the money to a card or immediately to purchase the desired product or service.
Of course, this seems like a fantasy for now. But AI technologies are developing so dynamically that all of the above could become a reality in just a few years.