The global technology industry needs to eliminate the AI bias by 2022
For a long time, technology has been promoted as ‘neutral’ and ‘non-partisan’. The dominant motto is “Neural is Neutral”, and over time it has evolved into “Virtual Is Neutral”. But nothing in the world is forever arbitrary. With the advent of the most sophisticated styles and brands of technology, there is a growing awareness of technological bias. Take the AI thing. Undoubtedly, the most sophisticated type is subject to constant criticism about its bias. Tech developers and promoters have unanimously responded to such criticisms and sought to counter them by arguing that AI could eliminate bias or at least reduce them by being in the ‘human loop’. Is it really so?
The main idea behind this phrase, ‘AI Bias’, despite the great advances in AI and the call for ‘AI autonomy’, is that there is a limit to where it can go and not just human intelligence and intelligence right here. Intervene and manage to gain the upper hand. To explore the point somewhat in depth, AI is inherently ‘schematically’ limited when humans are ‘organic’. Again, as time went on, a question arose, pointing to another turning point in the debate: Can ‘Human in the Loop’ really enable AI to get rid of bias?
While the human factor in managing AI is being propagated here, no one can underestimate the fact that there is also an opposite trend. Many leading experts in AI studies are confident that by the middle of this century AI will have grown exponentially from being a ‘supplement’ to the human brain to becoming a ‘guide’. Human thought and decision-making processes- be it in the political, economic or commercial domain. Since AI has the potential to reach new heights in ‘Super Intelligence’, the main point of the argument is that it can surpass human intelligence at any time. It refers not only to faster decisions but also to more ‘rational’, ‘objective’ and ‘accurate’ decision-making ecosystems. Since it comes from experts who have been seriously engaged in AI research for decades, such a claim cannot be completely ruled out and we call it ‘wrong’.
There is also an important issue of human understanding of AI when trying to rely on ‘Human in the Loop’ logic. It is a well known fact that AI moves fast and in many ways and its development including AI BIOS is not so easy to understand. The issue has become more complicated when it comes to AI applications as there is a misconception or mistrust among the users. This, in turn, leads to a number of legal, financial and ethical questions and issues that need to be addressed and negotiated successfully, not just by the ‘Human in the Loop’. The point to note here is that if AI Super Intelligence is fully implemented and users lag behind in understanding its functions, the day may come when AI-led decisions will take precedence over human-mediated decisions for general practical reasons.
AI does not need to be overly enthusiastic in predicting the specific time that humans are going to be overwhelmed. There are many adverse factors that AI may encounter, including the inability to recognize a specific context and respond accordingly. Also, AI is often hacked, which severely damages its credibility and autonomy. However, as the discussion reveals, keeping the ‘Human in the Loop’ strategy in a routine manner in an ultra-dynamic situation is not a viable solution.
So, for those who support the ‘Human in the Loop’ strategy, this is not a win-win situation. Not so for those who sing the tunes of AI’s ‘Unbound Autonomy’. In fact, with appropriate governance provisions in place to support the backup of AI Super Intelligence and Human Intelligence, there must be a search for the hitherto elusive correct point that forms a fair compound to provide a broader consumer interest.
Share this article
Do the work of sharing
About the author
More information about the author