Sloppy business practices: duck shoving thinking to inadequately programmed machine learning (“AI”)

An article on today’s The Guardian website examines the issue of bias on so-called “AI” - machine learning, to give it the more accurate and correct terminology (see https://www.theguardian.com/technology/2023/mar/27/robot-recruiters-can-bias-be-banished-from-ai-recruitment-hiring-artificial-intelligence).

This is far from the first time that bias in such systems have been raised. Problems have previously been attributed to a wide range of causes, such as unconscious bias in programmers (which is why such systems are NOT “blind”), and this article also examines a range of issues - including that the “learning” could be biased by such matters as the prevalence of male candidates, causing it to downgrade other candidates. 

There are, however, some issues that I consider should also be taken into consideration: 

• blatant bigotry, rather than unconscious bias, of programmers and all others involved in the creation, selection, and use of such systems; 
• bias against introversion; and
• time pressures on decision making in business.
  

I want to consider the third of those points further in this post. 

It’s easy enough to say “the decision making was flawed”, but why was it flawed? Typical assessment of that question looks at thinking styles, evidence, etc ... but ignores the context in which the decision making is being made. 

Weighing evidence and thinking in a lab is one thing: doing it in the clamour of multiple competing production deadlines and crises is another./ Under those circumstances, it is understandably human - and fatally flawed - to see something that promises to save time or take pressure off as a lifesaver - like a buoy thrown to a drowning swimmer. 

What to do about that problem - generally, not only in the context of considering AI systems? 

Acknowledge the importance of context, and ensure adequate decompressing, destressing and time commensurate with the importance of decision making. 

(This is something that politicians foolishly holding all-night sessions to rush decisions through should also consider, IMO.)

 

Assumptions / basis 

In writing this, I have assumed / started from the following: 

• this blog states quite clearly that it is about political and human rights matters, including lived experience of problems, and thus I will assume readers are reasonable people who have noted the content warning in the post header; 
• emotions are valid and important;
  

Possible flaws 

Where I can, I will try to highlight possible flaws / issues you should consider:

• there may be flawed logical arguments in the above: to find out more about such flaws and thinking generally, I recommend Brendan  Myers’ free online course “Clear and Present Thinking”; 
  
• I could be wrong - so keep your thinking caps on, and make up your own minds for yourself.
  

 

If you appreciated this post, please consider promoting it - there are some links below.

Finally, remember: we need to be more human being rather than human doing.