Category Archives: ai

An orbit of the sun

Posted on by

So a year in and a few things are apparent. First I am not really a blogger. While I have many ideas for interesting blog posts my focus is on the idea. The activity to turn it into a blog post worth publishing seems to add little value for me – so it doesn’t happen. The idea gets logged and I move on to the next.

I enjoy what I do. I had frequently thought to myself ‘in an ideal world what is it that I would do?’ and found that there were no evident opportunities out there to do it. For me it is to understand knowledge using the medium of ideas. There are opportunities once this has progressed to commercialise – but commercialisation is not a directing objective, and brings along with it its own negative trappings. Demanding a focus on the machinery of commercialisation rather than the underlying inspiring activity.

The cult of productivity is ingrained in the current commercial culture and seems to be becoming more ingrained. By exposure it also habituates into the minds of the participants in that culture. But an optimisation of productivity also digs a pit on the top of a hill that is a barrier to change when that hill is discovered to be a local maxima – or even inhibits the vision to see that the hill is a local maxima. For me this resulted in an initial focus on activities that, while they delivered visible progress, denied me the opportunity to deconstruct the foundational pillars they were built upon.

In the field of AI it is naive to think that we are even close to what is achievable. Many of the current generation solutions are little more than complex calculation tools that can solve a specific problem of a general class of problems. They can add significant value to what we do at the moment, so are not to be dismissed, but they are just a step along the way. Much as balancing one flat rock on top of another is a step along the way to reaching the current capabilities of architectural construction.

So far my focus has been on generating ideas towards reasonably undirected objectives.  A process that I have developed reasonably useful practices around. I will continue that. While I am not imposing direction on my future self, a couple of ideas that might indicate near term activities are; to try to project into code a paradigm that more closely fits with working with knowledge and ideas, and designing practices and expression forms that allow me to build second order value on top of ideas.

Playlist

Dancing / Kylie Minogue

The ossification of belief

Posted on by

Quite regularly my theoretical work on AI throws up or suggests ideas that are applicable to other areas of study. A detail in a structure I am investigating will spark across to another field and form an idea that is not directly relevant to what I am doing, but interesting none the less.

A recent example of this is the idea that beliefs in people become entrenched over time. Any belief, whether it be about the state of the world, a behaviour, a good way to do things or even muscle memory. When learning to type or changing a password how often do you find yourself making the same mistake each time, until it slowly gets unlearnt?

Beliefs seem to be most malleable when they are first formed. Before any reinforcement has strengthened and solidified the belief.  As such the best time to correct an incorrect belief is as soon as it is formed.

Some lifestyles tend to lend themselves to an easier engagement with modification of personal beliefs. Typically where incorrect beliefs get challenged and can be shown to be wrong in a way that the individual can accept. Research science is a prime example.

Where an individual encounters no challenges to their belief set, or the beliefs are of a form that have no obvious mechanism to prove them right or wrong, then an incorrect belief once formed is likely to linger for a long time.

Playlist

Under pressure / Queen and David Bowie