As AI develops, it will begin to affect every facet of our lives. Governments need to be ready.
In the UK, A-level and BTEC students have recently received their results. Due to the COVID-19 pandemic, however, things are different. The students’ grades have been determined by two things. Firstly, their predicted grades – given by their teachers and based off past exam performance, homework and personal student assessments. Secondly, the government exam regulator’s algorithm. This has opened the door to massive controversy as almost 40% of teacher’s predictions were reduced by said algorithm, hitting those from lower socioeconomic backgrounds the hardest.
Grade inflation statistics indicate that independent schools received a significantly higher boost to grades than comprehensive schools. Many critics from across the political spectrum have accused the government of widening the inequality gap in the UK. However, this may simply be the first time where the coming generation will face the mercy of algorithms.
As AI (artificial intelligence) becomes more sophisticated, it will begin creeping into our lives. In the coming years, one particular branch – decision making AI – has the potential to affect us in ways that were unthinkable 50 years ago.
Decision making AI is when computers replicate the cognitive decision-making abilities of humans. It is being explored by businesses to help make shrewd decisions by spotting market trends. However, it brings up several moral issues, with human rights groups beginning to express their concerns about how it is implemented in the future and who could be worst affected.
AI is already used by public offices in the UK for minor things, such as ensuring MOT standards remain high.
Controversially, in the USA, it is used by the criminal justice system. The AI-assisted software generates scores for offenders, quantifying the likelihood that they will re-offend to determine their bail. This is another example of computer algorithms having very real human impacts on our lives. The advocates for it claim that it can reduce backlogs of cases and free humans up to do more meaningful work.
However, critics are cautious due to the so-called ‘black-box’ effect. This is where technological complexities in the decision-making process mean that the specifics are hidden from humans. This leaves the choices being made open to massive – and invisible – bias.
In Los Angeles, they use predictive policing to anticipate crime and assign officers. The risks involved in this are that it relies on already-collected data that contains its own biases. Black and Latino communities will be affected the most, which just widens the very clear biases in the American policing system.
Recruitment is another area that will be changed due to AI. Employers (especially in the post-Coronavirus economy) can be sent thousands of applications in a very short period of time. Most companies can’t humanly process every single application or give it the attention it deserves. AI could help though, assisting employers to give every applicant a fighting chance.
Again, bias is a major hurdle to this.
In 2017, Amazon had to ditch its AI tool that they used to help filter out applicants. It was revealed that it penalised applications that used the word ‘woman’ in them. This was because the AI assessed the employment landscape of Amazon and found that there were more men than women, so effectively taught itself that men were preferable candidates to women.
This aptly demonstrates the risk that happen when using AI-assisted tools. They are just as likely to be biased, if not carefully considered.
The simple fact is that AI is inevitable. Nothing can put it off. Therefore, instead of fighting against it, it may be prudent to ensure that laws are in place to make it more transparent and less vulnerable to bias.
If we are careful, it will be able to do massive amounts of good and potentially eliminate human bias. This is because AI has no personal biases in the same way as humans, AI has no race, gender or sexuality. If it is deployed properly, it could help the journey to a more egalitarian world.
The A-level fiasco is one of the first widely publicised times where an algorithm has demonstrated bias and unfairly affected certain members of society. Unfortunately, however, it will definitely not be the last.
Governments across the world should take this as a wake-up call to the changing world. We need rules imposed to make sure AI will help to close the gaps in society, not widen them.