Torque Clustering: “Autonomous AI on the horizon” – A new algorithm significantly improves how AI can independently learn and uncover patterns in data.

From the article

>Torque Clustering can efficiently and autonomously analyse vast amounts of data in fields such as biology, chemistry, astronomy, psychology, finance and medicine, revealing new insights such as detecting disease patterns, uncovering fraud, or understanding behaviour.

>“In nature, animals learn by observing, exploring, and interacting with their environment, without explicit instructions. The next wave of AI, ‘unsupervised learning’ aims to mimic this approach,” said [Distinguished Professor CT Lin](https://profiles.uts.edu.au/Chin-Teng.Lin) from the University of Technology Sydney (UTS).

>“Nearly all current AI technologies rely on ‘supervised learning’, an AI training method that requires large amounts of data to be labelled by a human using predefined categories or values, so that the AI can make predictions and see relationships.

>“Supervised learning has a number of limitations. Labelling data is costly, time-consuming and often impractical for complex or large-scale tasks. Unsupervised learning, by contrast, works without labelled data, uncovering the inherent structures and patterns within datasets.”

I have to laugh when I see people basing their ideas on how AI will play out on nothing but current state of the art or even just current LLMs

The kicker is, every advance in fundamentals technique like this makes the next set of advances in it much easier, the scaling potential is enormous. There is likely to be some sort of bottleneck somewhere that plateaus it out but its clear we are no where near that.

Aside from space I barely think very much about what may happen beyond 2050 at the minute, technology is likely to change so much by then that theres barely any solid basis to speculate from beyond that.

One relatively minor and mundane consequence for example, domestic miniaturised AI systems with 20 years of advancement at the current pace effectively spell the end of mass media entertainment as we know it. You can just ask for anything you want to have that a TV or potentially a VR set can do. Disney have holodeck style moving floors right now so potentially you can throw in the domestic version of that too. Whole franchises / universes will be created with little Human intervention and shared a la the SCP Foundation (collaborative x-files project, worth reading) and at practically zero production cost.

It sounds very heady but this is just a relatively small consequence of this giant leap in abilities we are currently in thats not even particularly speculative, most of it is current tech + business as usual AI development rates. In the future we will be combining this with technologies no one has even thought of yet and combining things in unforeseen ways.

Torque clustering itself indicates this is the correct expectation, it creates a pretty major efficiency gain which will lower the barrier to entry (greatly reduced labour requirement – suddenly specialist SME’s are entering the frame) just as advancing computer chips eventually created the home PC.

Independent and supervised learning has been a goal for a while. However a major issue facing it (and there are many) is the problem of alignment; making sure AI understands what we want to achieve and having it perform actions that are consistent with the general value systems of our civilization. Obviously that can be a lot of different things depending on the country and political context but even something that appears simple – dont cause the extinction of humanity – can be deceptively difficult to get AI alignment on

Its the whole 3 wishes from a genie scenario – no wish can ever be explicit enough to remove any and all interpretations that lead to generally misaligned or potentially harmful outcomes.

Friendly format [here](https://youtu.be/gpBqw2sTD08)

Deeper dives [here](https://m.youtube.com/@RobertMilesAI)

*This is not an ad, I just really like these channels