The Rise of The Planet of AI

With this algorithm, we have been able to implement it in a variety of our daily tech encounters. AI sits behind the driving force of Google Maps and navigation, as well as voice control. It can easily be found anywhere. “Just because these examples do not exhibit super intelligence, does not mean they are not useful,” Dr Sonenberg explains. 

Dr Sonenberg alludes that the real problem with AI lies within it’s fairness. These “pattern generating algorithms” at times “learn from training sets not representative of the whole population, which can end up with biased answers.” With a flawed training set, a flawed system is in place. This can be harmful to certain demographics and cause a sway on consumer habits. 

With AI-aided advice, the explanation behind outcomes and decisions are not supported either. Algorithms are only able to mechanically produce an output, but not explain them. With more high-stakes decisions untrusted upon the reliability of AI, the issue of flawed algorithms becomes more pronounced. 

With my interview with Dr Sonenberg, not one moment was the fear of super-intelligence, robot uprisings, and the likes brought up...

With the new-found knowledge of AI’s current concerns I brought up with Dr Sonenberg, I conducted another interview with Dr Tim Miller, a Professor of Computer Science in the School of Computing and Information Systems at The University of Melbourne, and Dr Jeannie Paterson, a Professor teaching subjects in law and emerging technologies in the School of Law at The University of Melbourne. They both are also Co-Directors at The Centre for Artificial Intelligence and Digital Ethics (CAIDE). 

As we began the interview, Dr Miller explained again that AI “is not magic” and implements the use of “math and statistics”.

Dr Paterson was clear to bring up that anti-discrimination laws have been in place but as technology evolves and embeds itself more into public domain, it must be scrutinised. The deployment of AI can easily cause harm to people due to systems not being public, causing sources to be difficult to identify and causily attribute.

With the prospect of biased algorithms, a fine dissonance occurs. Dr Miller elaborated on the use of AI in medical imaging used in private hospitals. As private hospitals tend to attract a certain echelon of society, the training set is not wholly representative of the greater population. “A dilemma occurs with racist algorithms… if it is not used [outcomes] could be worse.”

When the idea of a potential super-intelligent robot emerging in the future was brought into conversation, the two didn’t seem to be very impressed. 

“Don’t attribute superhuman qualities [to it],” says Dr Paterson.

Dr Miller states that the trajectory of AI’s future is difficult to map. Predictions in the past of how AI progresses with it’s abilities have occurred, but they occur much later than expected… easily decades later. The idea of super-intelligence also poses the question on how to define intelligence. “Intelligence is multidimensional, it has its limits,” says Dr Miller. 

In this mystical future world of AI, a distinction is placed not just on, “what will machines be able to do but what will not have them do,” states Dr Miller. “This regards anything that requires social interaction, creativity and leadership”; so the future is aided by AI, not dictated by it. 

However, in a more near future, some very real concerns are posed. Job security, influence on consumer habits, transparency, law approach, and accountability are only a few.

With more and more jobs being replaced by machines, every industry is at stake. “Anything repetitive can be automated,” says Dr Miller. But this does not instinctively pose a negative, as more jobs will be created to further aid the use of AI. And not all functions of a job can be replaced by AI. Dr Paterson explains with the example of radiology that AI is able to diagnose and interpret scans, but a radiologist does more than just diagnose and interpret on a daily basis. “The AI is used to aid in the already existing profession, not simply overtake it.”

Greater transparency is needed in showing how AI uses our data. “It shouldn’t be used to collect data unlimitedly,” says Dr Paterson, “is it doing what’s being promised, is it discriminating people, is it embedding inequality?” With this in mind, Dr Paterson suggests that more law authorities should be educated on how to approach topics regarding AI. “There needs [to be] better explanation… [We] need to educate judges and lawyers.”

With the notorious Facebook-Cambridge Analytica scandal of 2018, the big question of accountability was raised. The scandal involved the unwarranted use of data from 87 million Facebook users by Cambridge Analytica which served to support the Trump campaign. This scandal brought to light how the data we used can be exploited nonconsensually and used to influence our behaviours, as this particular example seemed to sway the American presidential election. 

Simply put, our information can be easily exploited and sent off to data analytics to further influence our choices. This creates the defence that apps “ merely provide a [service], but people use [these services] in that way,” as said by Dr Miller. Simply put, the blame becomes falsely shifted onto the users for the spread of misinformation. The impetus, however, should lie with social networking sites disclosing to it’s users more transparency on their data usage and history as well as providing adequate protection on their data.

To be frank, the future of robotic humanoid AI integrating seamlessly into human livelihoods will not occur within our lifetimes, or potentially even our grandchildren’s. The forecast seems at best, unpredictable; and at worst, unattainable due to the complexity of what constitutes full “sentience”. However, this does not indicate that AI lies dormant within our lives. The fundamental technology based in computing, statistics, and information systems lays most of the groundwork for most transactions we conduct online, whether monetary or social or otherwise. AI and it’s promises should not be shunted aside due to the misleading media surrounding it’s popularised definition and “robot uprisings” but rather taught more broadly to all audiences. 

So perhaps Elon Musk’s fantastical ideas of robotic integration will not occur by 2022 but the presence of AI in modern technologies should not go unnoticed.