Artificial Intelligence (AI): Friend or Foe?
AI has undoubtedly transformed the way humans operate, but will the risks associated begin to outweigh the benefits reaped?
What was once limited to being a common theme within the realm of science fiction, has over time become a technology intertwined with our everyday lives in a way that is almost invisible and inconspicuous. From Amazon’s Alexa providing personal assistance within the home, to chatbots online providing solutions to our online shopping conundrums, AI has permeated our lives within countless spaces. Despite a report in 2016 predicting the 2020 global AI market to reach $5.05bn, in 2021, the market was worth a whopping $327.5bn. This demonstrates the huge growth which the sector has seen, and it would be wrong to assume that this is halting any time soon.
AI is a technology that is used to mimic the capabilities of human thinking, by training them with experiences which the machine can then utilise to understand and make efficient responses to language, decisions and problems. Industries such as high tech and telecommunications, financial services and healthcare and pharmaceutical are leading in applying AI within their businesses. However, AI can also be seen within other spaces such as agriculture, seen locally in Northeast England — where robots are being trained to classify the ripeness of strawberries without the risk of damaging them and the potential to scan up to 4 million a day!
Clearly then, AI poses the potential to play a pivotal role in shaping the workplace of future generations, maximising the efficiency of various processes like AI analytics, uncovering patterns from large datasets whilst eliminating the issues associated with human error. From this view, AI is presented as a technology that can facilitate the work of humans, reducing the time needed to carry out various tasks whilst allowing attention to be given to other ones, maximising the productivity of whole systems. Therefore, instead of being seen as a replacement to humans, the benefits of AI are to act as an aid to the work humans are already doing.
Amazon have also even attempted to apply AI within their recruitment processes, but has this been successful? In summary, no. What really happened was that the inequalities and biases present within the ‘real world’ so to speak, translated into the screening of applications, disadvantaging women. This is because, the mechanism was based on reviewing the patterns in the resumes of Amazon employees over the last 10 years and lo and behold, given the dominance of men in the Tech industry, the tool taught itself that male candidates were preferable.
Unfortunately, these biases in recruitment and AI are not limited to gender related ones either, with anti-Black biases found too. Academic recommendations made by job platforms for 63% of respondents (black women) “were lower than their current academic achievements”, which is particularly concerning given that black women are the most educated group in America. Clearly, the role AI has within hiring is still limited and must not be implemented without further consideration for how they can be equitable and unbiased, instead of reproducing inequalities. Within the academic space, the erasure of significant technological contributions by black women and other minorities is still something that needs to be addressed, alongside improving the industry culture.
There are however, efforts being made to tackle these issues, with Spotify looking to expand their AI ethics team, with the aim of assessing their algorithms’ impact, since a 2021 study demonstrated that music recommendation algorithms were more likely to promote male artists. Rob at PwC also emphasises the importance of having diverse groups being directly involved with the designing, training and deployment of AI, in order to take into regard the ethical implications of such technology. In addition, he speaks of the increasingly interdisciplinary work between professionals occurring when solving these multifaceted problems.
However, data privacy concerns remain in the AI sphere. These can involve data persistence whereby data exists longer than the humans that created it, data spillovers where victims have no voice given they are unaware of their involvement and data repurposing. This is an issue given the breaches, fraud and identity theft that can occur when data is in the wrong hands, and companies can lose out on $3.86 million per breach. But —solutions are being created by Ship2MyID to separate the Personality Identifiable Information (PII) of consumers from their behaviour related data, to protect them from being exploited by businesses trying to sell their products.
This links to the idea of having an increasingly offline presence, whereby devices like Echo or Google Home are collecting our data (since microphones remain on) even when we are not directly engaging with “online” activities. This highlights the expansion of AI within both mainstream society and our homes, and the dependence formed to fulfil everyday activities. This increasing surveillance can be of concern, but others argue we hand over our data voluntarily in exchange for “free” convenient services. Overall though, consumers globally are six times (45%) more likely to view AI’s effects on society as positive rather than negative (7%), as well as within their personal lives (52% vs 7% correspondingly).
Similarly, previous students have shared their views on our future with AI with Angela adamant that AI is limited to the instructions given and therefore the idea of them replacing humans is extremely unlikely. This contrasts with speculations that upcoming decades may see Artificial General Intelligence (AGI) undertake “recursive self improvement”, developing “basic AI drives”, ultimately seeing humans as a threat to their self determination, causing them to pursue human extinction. However, containment solutions are underway though it is argued that long term strategies focused on value learning and corrigibility — ensuring human values are adhered to and systems are open to being corrected — are essential.
It’s clear that issues of deceit and trust are present when consuming AI technology — particularly with anthropomorphism whereby robots are becoming more and more life like. Shifting negative public perceptions of AI will therefore only occur if bots are clearly labelled and service providers are transparent with their consumers. In addition, inequalities within the industry and technology must be addressed to avoid further biases being replicated. Optimistically though, AI could be viewed as facilitating, providing solutions to our problems. Regulations and policies such as the UK’s AI sector deal present efforts to progress AI development, with people at its core to avoid AI from becoming a definitive Foe.
