“Facebook’s experimental chatbots talk to themselves in secret languages!”
“Self-driving cars are racist because they’re more likely to run over black people!”
“Is a robot coming for your job?”
There are plenty of AI scare stories out there. And they make for great headlines. But great headlines often ignore the nuances and realities of a situation. So what’s the truth?
Well: artificial intelligences aren’t plotting behind our backs. They just simplify conversations by communicating ideas without the arbitrary rules of grammar, which happens to make their output harder for human users to read. It’s about as dangerous as textspeak. Self-driving cars’ video cameras do have trouble with dark colours – that’s why they have a backup steering system that uses laser light pulses to detect pedestrians, regardless of the colour they happen to be.
And AIs aren’t out to steal your jobs. In fact, quite the opposite. While automation will undoubtedly disrupt many industries, it will simply change the way we work. And a lot of times for the better. So let’s look at the reality; the flipside. What are the positive by-products of AI, and how will it improve recruitment, the world of work and your job in general?
AI in recruitment
If you believe the horror stories, human resources and recruitment personnel will be up against the wall when AI takes over. The recruitment process involves processing an enormous amount of data. The processing power of an AI algorithm can scan, sort and analyse this data at high speed, benefitting recruiters and candidates alike. AI can detect biased language in job descriptions, making them more attractive to marginalised or underrepresented demographics. Chatbots can make initial contact with potential candidates, ask screening questions and assess the answers according to the rubric, with no subconscious biases in the moment.
“Cutting down AI bias is a belt-and-braces affair.”
AI systems do need to be set up with the right data, to avoid them learning systemic biases from the datasets used to train them. This is why researchers at the University of California have worked up a machine learning scheme that identifies and favours gender neutral terms. It’s also why MIT’s Computer Science Lab recommends altering samples to correct for bias (increasing the representation of marginalised groups so the AI has more material to learn from). Cutting down AI bias is a belt-and-braces affair, which involves putting people to work on properly training their AI helpmates, as well as improving the design of the AI itself.
It can’t be denied that AIs will take on some work, replacing some jobs – including a predicted 16% of jobs in recruitment by 2029. This isn’t so much a question of replacing people, however. The key is to think of “job” meaning not “role or vacancy”, but “task”. The mundane tasks of sourcing, screening and matching candidates can be carried out by “narrow AI” tools – that is, tools which solve certain, specific problems with superhuman accuracy, but aren’t anything close to the human-like intelligence of “general AI”, which is still decades away at best.
By automating tiresome, time-consuming tasks in the workplace, AI will free up the working hours of employees to focus on person-to-person interactions and decision making – like the essential job of eliminating unwanted bias in AI systems.
AI in work
A 2018 report by the World Economic Forum claims that by 2025, more than half of all current workplace tasks will be carried out by machines – but this doesn’t mean that more than half the roles for humans will have simply ceased to exist. Quite the opposite, in fact; the same research predicts that while around 75 million jobs will be displaced, 133 million new ones will be created.
As research from the MIT Sloan School of Management puts it, AI will not replace jobs so much as redesign them – which means the matter at hand for candidates is not “losing my job to AI” but “reskilling and upskilling for my new job” – whether that’s a change of career, or a change in the tasks your existing job demands of you.
Programmers may find themselves with different responsibilities in the workplace, as AI takes over the donkey work of producing and testing code. The same skill set is likely to be reapplied from producing code to checking it. Data scientists and intelligence analysts will need to confirm and apply the findings of the AI workmates. And cybersecurity professionals will still be needed to oversee and sign off automated actions to avoid errors and oversights compounding themselves.
In other roles, AI will take on a collaborative function – it’ll be a new tool for professionals to learn so they can improve the work they already do.
Radiologists, for example, might use AI to read medical images in more depth, but they’ll still have to use their existing and informed judgement to interpret what the AI helps them notice. Hands-on, rote management functions will likely fade away, but the new roles will involve skills AI doesn’t have like intuition and creativity. The work that is left for humans to do will be the fulfilling, rewarding kind of work that we actually want to do.
Added benefits of AI for workers
Writing for the Foundation for Economic Education, Don Lim observes that automation frees up time and opens the possibility for multitasking, working secondary roles or on passion projects: we may yet see the return of Genius Hour to the tech sector, allowing everyone time to innovate, develop and grow on company time.
“AI is also likely to improve morale in the workplace for everyone.”
If a move to the creative industries doesn’t appeal, take heart; AI is also likely to improve morale in the workplace for everyone. Besides eliminating the drudge tasks, AI can show workers exactly how their work contributes, providing unbiased performance reviews and even adjusting schedules to match predicted demand. AI can directly improve workplace wellbeing, too, by helping to identify employees who are in need of specific interventions, and particular times of year when stress runs high and care policies need to be implemented.
AI tools can also improve accessibility in the workplace, using biometric logins to remove the barriers people with cognitive or lexical difficulties have in remembering and typing passwords. Salesforce is working on an algorithm that automatically summarises text, improving access to information for people with cognition, attention or literacy impairments. Amazon’s Echo already allows blind or visually impaired users to interact with the web without eyes (or hands, for that matter).
Make no mistake: artificial intelligence could radically change the entire structure of our work-to-live society, eliminating the need to work for survival and making universal basic income a necessity. But AI won’t totally replace every working human being everywhere with a machine. The narrow AI that’s currently being used in hospitals, factories and offices around the world is only capable of helping human beings by processing information, finding details, and easing interactions with other technologies. It can’t make decisions, or come up with ideas; but it can, and will, make our working lives more productive and fulfilling.
AI also powers our predictive psychometric test – the tool we use to match candidates to the best career opportunities and make the right recommendations for us and you. Why not give it a try?
