When Google’s DeepMind AI beat the world AlphaGo champion Lee Sedol in March 2016, a shock wave ran through society as a stark reminder of the rapid and inevitable evolution of AI.
Not long ago, Musk stated that Artificial Intelligence posed a “fundamental risk to the existence of civilization.” His apocalyptic argument touched upon the real risk of humans becoming second-class citizens in a Terminator-style world where robots have the upper hand.
Musk reached far into the future when he treated the topic of neuroprosthetics, enhancing cognitive and sensory abilities, enabling us to communicate telepathically.
Although many may dismiss Musk as a fearmongering doomsayer, “mind-controlled” drone racing and an ability to “feel” a prosthetic hand indicate that he may in fact have a point.
In practical terms, Musk is calling for proactive regulations. And before we embark on further experimentation, there is reason to believe that we can benefit from assembling a protocol to guide future action.
To say that Zuckerberg disagrees with Musk would be a colossal understatement. Zuckerberg went so far as to declare the naysayers and doomsday predictors downright “irresponsible”.
The Facebook CEO essentially highlighted AI as a force for good, providing humanity with unprecedented means to eradicate hitherto indestructible problems.
Zuckerberg’s own aim to expand his knowledge of AI counts Jarvis, an AI assistant who handles many daily chores such as opening doors for people it recognises and entertaining his 1-year old toddler. There is a light-hearted mood surrounding Zuckerberg’s interaction with the virtual assistant voiced by Morgan Freeman, and it allows us to sympathise with his positive interpretations of the future of AI.
At the same time, there is reason to believe that Jarvis wasn’t the culprit in Musk’s predictions. Rather, it was the offspring of advanced intelligence that troubled the Tesla man.
Hawking and Omohundro
Bill Gates and Stephen Hawking joined hands with Elon Musk on the Doomsday predictions in an open letter encouraging a fundamental assessment of the pitfalls of AI.
Hawking asked about the extent to which we believe that the intelligence of the robot can in fact surpass that of its creator. Can AI generate its own progeny without limitations to its mental capacities? Regardless of the fictiveness of Ex-Machina, the film brought matters to a head by showing us what happens when the emotional abilities of manipulation and calculation allow a robot to outsmart its creator.
In his paper The Basic AI Drives, AI scientist, Steve Omohundro explains that to circumvent the realisations of Musk’s gloomy predictions, we need to be exceptionally careful in how we go about designing our robots. He argues that the inherent “drives” of the robot will lead it to resist human control and copy itself without concern for its surroundings. It will do so not because of programming errors but due to the intrinsic nature of goal driven systems. A compelling argument from an expert and president of Self-aware Systems which builds AI capable of self-reflection and in possession of a desire to improve.
The Societal Debate
In the current day, one focal point of the AI debate has been that of employment. The camps split more or less straight down the middle, and the arguments have rotated around whether robots would replace human labour or add jobs to the market.
At the World Government Summit in Dubai in 2017, Elon Musk declared AI capable of triggering irreversible mass unemployment. His argument finds a great deal of historic endorsement since automation has already robbed millions of blue collar workers of their jobs across the globe.
The counterargument to the mass unemployment dystopia dwells at the self-regulatory abilities of the market forces. According to the neoliberal logic, the supply-demand equation would render mass unemployment impossible because, at the end of the day, without consumers with fat wallets who would we sell to?
But perhaps that argument is slightly too black and white. Since the 1950s, the world has witnessed a highly problematic bifurcation of the workforce. In the US, automation laid off thousands of blue collar workers whose factory jobs were either in China or in the hands of a machine.
What ensued was decades of unemployment so widespread, it created what today is known as the underclass. Those are the chronically jobless welfare recipients whose faith has been widely debated both politically and academically. So evidently, automation has a tendency to socioeconomically polarise societies and deepen inequality.
However, Larry Elliot emphasised how the nature of the job had everything to say about the likelihood of eliminating its own existence. Logic and algebra, he argued, were characteristic of middle-class jobs and more likely to be replaced by AI whereas tasks requiring perception and mobility were much less so. A simple reflection of a robot’s elasticity.
In many areas, we are already detecting negative discrepancies between employment rates and earnings. Some may blame the neoliberal crusade and although they may be right, tech and politics are often inextricably linked. For that reason, we may have to rethink our political system all together and reconsider what it can tackle and handle- and especially what it cannot.
But as we debate the possibility of robot world domination, we need to debate the intelligence element of AI. A multi-faceted, multi-layered capacity, the strength of its ethical and moral compass goes back to Omohundro’s argument and will depend entirely on how we decide to construct the artificiality. The fact that robots have replicated the sexist and racist bias humans possess tell us a great deal about the influence we as humans have on the AI state of mind. The Twitter-trending #maketechhuman hashtag accurately encapsulates this concern.
The Terminator Effect
With the perils gathering all the attention, the positives also deserve their share of scrutiny. Maybe we just watched too many films. Ironic in the context of the AI debate is the fact that the vast majority of AI creators has benevolent agendas to scrap the inconveniences of daily life whether logistically or in terms of health.
Imagine getting from your living room to the office in 3 minutes and 39 seconds or no longer battling cancer or HIV. Zuckerberg’s Jarvis may have been for sheer comfort but the virtual assistant’s technical IQ suggests that maybe he would be able to carry out more profound tasks that could help beyond Zuckerberg’s immediate family.
Gecko Systems built CareBot whose job it is to cater for the elderly and the disabled. Hardly anything harmful about that. Hoovering up dust bunnies without a pair of human hands also hasn’t caused much outrage.
In short, the status quo is manageable. AI is currently stronger on the artificial than on the intelligence and that is comforting. It hands credit to the positive wing.
AI and Human Rights
At the very core of the debate lies the question of whether we can in fact call Artificial Intelligence intelligent. Where do we tolerate the location of the final frontier?
The EU began seriously contemplating whether we owed AI a technical equivalent of human rights, granting “electronic personhood” to safeguard rights and responsibilities. The EU’s reasoning behind this proposal was to ensure that robots do in fact “remain in the service of humans.”
Europe’s talks suggested that the ramifications stretch far into the political and socioeconomic fabric of our societies and, however inadvertently, aligned with Musk on the prognosis of human existence on the periphery of life. Humanity is in dire need of assembling a framework of legal definitions and codes of conduct.
It is the prospect of self-evolving software that instills a sense of urgency that AI could go rogue anytime. For many, the fear may emanate from an inability to pinpoint where we are on the spectrum ranging from tagging cucumbers in pictures and curating newsfeeds to summering on Mars and building weapons of mass destruction. Is this the last junction we will reach before annihilating the human race?
War of Words
In an unfortunate propensity towards a right and wrong mentality, the seriousness of the issue and its possible ramifications call for a discourse that suppresses the propensity towards the right and wrong mentality currently skewing the discussion. It gets in the way of a well-informed assessment of the pros and cons that will ultimately prevent AI from running amok. Escaping human obsolescence means trusting that when robots are left to their own devices, they will disarm bombs rather than firing missiles.
It doesn’t matter who’s wrong or right- we have to make sure it’s Zuckerberg by taking the precautions Musk and Hawking have handed us.