‘AI’ or indeed any model or software should only ever be used to support decision making.

See the blog on ‘AI does not think’ for more information on the development cycle and why models are not sophisticated.

The model or ‘AI’ is bounded in that it can only perform processes that it has been told to do with very specific programming by a human.

The ‘AI’ or model can’t think it can only don as it is told. In addition to enable the model to perform multiple functions each one has to be programmed specifically and work alongside other units of code to ensure a joined up overall process. The caveat to this is that if you join certain decision making processes up and amalgamate data the outcome could be catastrophic in terms of societal control or inappropriate decision making.

This is the serious risk in automating human decision making and why, in many cases, it is inappropriate.

Watch the FRACTAI tik tok on this