Should we fear or embrace the future?

AI has unleashed a similar mix of utopian rhapsodies and alarmist anxieties as when TV and the internet were in their ­infancy. Will it improve, or hinder, our lives?

Will AI improve, or hinder, our lives?

Ever since Socrates feared writing would atrophy his memory and so refused to take up the newfangled invention, humans have blamed technology for their degeneration. With data now being the world’s largest business, the fast-accelerating field of AI has unleashed a similar mix of utopian rhapsodies and alarmist anxieties as when TV and the internet were in their ­infancy.

In I Human, organisational psychologist Tomas Chamorro-Premuzic avoids taking sides and instead analyses the impact of AI on our humanity, exploring the potential to harness algorithmic systems to improve ourselves while managing their risks. He acknowledges that personalised social media feeds perpetuate self-absorption and hinder deep thinking, making us more prone to distraction, bias, impulsiveness, and narcissism. His response, however, is not to reject AI but to demand more from it, and use data to identify our errors, build self-awareness, and ultimately better ourselves.

But while popular approaches to psychology sometimes claim that human behaviour is easy to change, Chamorro-Premuzic is less optimistic. He urges humans to train and harness AI as a corrective tool to compensate for our shortcomings. AI has no skin in the game; its power lies in its neutrality and objectivity.

The “be happy” mantra of positive psychology, which emerged partly as a response to the dehumanising impact of tech efficiency, risks optimising for personal satisfaction over genuine accomplishment or human progress.

These approaches can exacerbate the self-regard fuelling the status anxiety and quest for external validation in our digital lives. Chamorro-Premuzic is similarly acerbic about the rise of “cognitive bias” training in workplaces, arguing these programs have no impact on behaviour; they cater to an audience that already considers themselves open-minded, while pushing us to blame others for biases rather than ourselves or the system.

I Human: AI, Automation, and the Quest to Reclaim What Makes Us Unique 

He suggests AI can more effectively enhance diversity and inclusion by using it to minimise biases in recruitment. His vision is for humans to become more involved in curating the data to train AI, transforming it into a tool to expose and overcome our blind spots.

Chamorro-Premuzic’s optimism is refreshing, but his articulation of how AI can improve us doesn’t fully land. It is more difficult to remain broad-minded when algorithms serve information that confirms our existing attitudes and inundate us with distractions that impede the reflection necessary for seeking different perspectives. Sensory overstimulation comes with intellectual understimulation and Chamorro-Premuzic cites evidence that multi-tasking is twice as mentally debilitating as smoking marijuana.

AI and automation have the potential to free us up for more creative and higher-order tasks, but how are most of us reinvesting these mental resources? We can train ourselves to become more in control of our browsing behaviour and resist the temptations of algorithmically recommended distractions, but the effort involved in self-restraint may also deplete our energy reserves.

Now that knowledge is so readily available, he encourages leaders to use AI to focus more on understanding the full complexity of scenarios, asking questions in a way machines can’t, and evaluating the quality of answers.

“In the AI age, the essence of human intelligence is highly conflated with humility and curiosity, perhaps more so than with knowledge,” he writes. But that the AI age will incentivise us to be more humble and inquisitive feels aspirational rather than an ­indication of progress under way.

■■■

Chamorro-Premuzic’s perspective on unbiased AI diverges starkly from that of New York University academic Meridith Broussard. She would likely tag his approach as “technochauvism”, her term throughout More Than A Glitch to describe what she considers is the prevailing social bias towards computational solutions over human-created ones.

Broussard rightly emphasises that many tech innovations are developed by relatively homogenous teams of young white men solving problems for themselves. This leads to the embedding of biases in tech, making them structural rather than mere “glitches” that can be fixed: “Digital technology is wonderful and world-changing: it is also racist, sexist, and ableist.” But her ongoing characterisation of tech warrants another label: “anthropomorphism”. AI itself is neither inherently good nor bad, but simply detects patterns and responds to the data provided to it.

Broussard presents powerful stories illustrating the danger of algorithms amplifying social inequities if left unchecked. Sensor technology, for example, is more effective on light than dark skin, sometimes leading to black people being mistakenly pursued for crimes they didn’t commit. Staggeringly, until last year, the NFL used race-based algorithms to calculate payouts to players who developed complications from brain injuries, assuming that black former players started with inferior cognitive function.

Unfortunately, Broussard’s important messages get lost amid her unwavering cynicism about tech and cherrypicked anecdotes. Sure, the Honolulu Police Department were overzealous investing $150,000 USD in a robotic dog following the outbreak of Covid-19. “Spot” monitored the temperature of homeless people from a distance of eight feet, but was unable to operate in rain or snow. However, an argument against investing in policing tech, this is not.

It’s surprising how slowly voice assistants have evolved to better serve stutterers, but Broussard fails to mention the obvious accessibility advantages of this tech to the visually impaired. Flip comments like “AI is not environmentally responsible” don’t ­bolster her credibility. Many companies are developing more ­energy-efficient ways to train AI models and myriad disrupters are innovating with AI to reduce carbon emissions.

Her catalogue of shocking algorithmically induced harms highlights the importance of training machines with diverse datasets and implementing human checks when AI recommendations have significant human consequences. But even with humans interpreting computational results, Broussard remains sceptical: “The human checker may believe technology holds all the answers and that the system can’t possibly be wrong.”

More than a Glitch: Confronting Race, Gender, and Ability Bias in Tech 

Broussard chronicles numerous instances of hubristic computer scientists disregarding experts by claiming new technologies will upend existing fields. A prominent researcher like Google’s Geoffrey Hinton should have known better than to proclaim in 2017 that “it’s just completely obvious that in five years deep learning is going to do better than radiologists”.

But Broussard misses the point that in any advancing field, many innovations will be discarded or require iterative improvements, leading to assertions like: “Technochauvism is what led to the thousands of abandoned apps and defunct websites and failed platforms that litter our collective digital history.” At times she displays the lack of empathy of the tech enthusiasts she rails against, pinning the increase in shoplifting in the US to “cost-cutting attempts” of retailers installing self-checkout stations.

Broussard acknowledges that tech can be positively world-changing, noting that: “Sometimes the right tool is a computer. Sometimes it’s not.” But too often her argument comes across as an anti-tech screed.

She doesn’t touch on solutions until the final chapter, where we learn about artistic movements offering alternate visions (“Indigenous futurism, Afrofuturism, Africanfuturism, Desi futurism, Arab futurism, Asian futurism, South Asian futurism and Chicanafuturism are all gaining visibility”) but little about how tech companies are improving their products or how organisations are deploying AI to help level the social playing field.

Particularly unhelpful is her digression into the genealogy of the statistics field. She references three “outspoken racists and eugenicists who inspired Hitler” before approvingly quoting one sociologist that “current statistical methodologies were developed as part of the eugenics movement and continue to reflect the racist ideologies”.

For all her ostensible progressivism, a genealogist might identify a different kind of intellectual ancestor for Broussard: a 16th-century activist fearing that the invention of newspaper publishing would end society by eliminating the need for people to meet.

Ben Naparstek is a growth and marketing adviser for technology companies

I Human: AI, Automation, and the Quest to Reclaim What Makes Us Unique
By Tomas Chamorro-Premuzic
Harvard Business Review Press, Nonfiction
208pp, $49.99

More than a Glitch: Confronting Race, Gender, and Ability Bias in Tech
By Meredith Broussard
MIT Press, Nonfiction
248pp, $54.99