Saturday 28 January 2023

AI/ML risks and opportunities

I am currently researching topics such as 'information risk management' and 'risk treatment', using Google as usual and comparing it against ChatGPT to see how it fares. 

Many of the pages suggested by Google are either too superficial or (to varying extents) factually incorrect, biased or misguided, hence are of little to no value - in my personal opinion anyway, knowing that I am naturally subject to confirmation bias and others. Roughly three quarters of the pieces offered by Google are not worth opening, and a fair proportion of the remaining quarter are superficial marketing collateral (a.k.a. tripe).



If AI/ML robots such as ChatGPT simply suck up, blend and spew out such information indiscriminately in generating their outputs, regardless of the quality of their sources, they are likely to present and further the prevaling opinions, feeding the mindless group-think that spreads misinformation, conspiracy theories and so on, not just tripe. 

This situation highlights, for me, three key things:
  1. Critical thinking (such as thoroughly researching, sifting through considering, evaluating and validating information before drawing conclusions and making significant decisions based upon it) is an extremely important skill for this information age, one that large sections of the population apparently lack, hinting at a widespread systemic failure of our educational systems.

  2. The true knowledge, expertise, credibility and personal integrity of authors (both people and robots!) is unfortunately of limited value given the previous point. It doesn't really matter to most readers, particularly those who don't even consider it.


No comments:

Post a Comment

The floor is yours ...