AI replaceability fears bring an urgent need for people skills but not for the reasons you think
Machines cannot feel compassion—at most, they can mimic it (primarily through advances in spoken language technology and Natural Language Processing). This highlights the importance of developing advanced people skills and communication skills.
In a recent keynote presentation at IMI’s National Leadership Conference, IBM Ireland General Manager, Deborah Threadgold, shared insights around the profound impact of AI in the workplace.
Drawing on IBM’s own digital transformation journey, Threadgold noted how the integration of self-service solutions and AI technologies like chatbots has allowed IBM to streamline HR processes, such as identifying gaps in employee readiness for promotions or handling out-of-policy expense claims—tasks previously conducted manually at great time expense. Crucially, AI serves in this example as a tool to assist human decisionmakers, who make the final decision but are freed from countless hours of repetitive data gathering and paperwork to focus on other areas.
Conversations around AI tend to go one of two ways: on one hand, it has the potential to free people up from repetitive tasks so they can work on more sophisticated and intellectually stimulating tasks. On the other, fears around job loss, replaceability and inequality at the wider societal level. While the latter fears should not be dismissed and the former can sometimes veer towards naïve optimism, IBM offers a real-life example of the potential for AI to augment our capabilities rather than replace us.
But fears around replaceability are not going away, and why should they? As groundbreaking as IBM’s transformation might be in terms of Future of Work potential, the positive impacts are currently reaped by skilled, educated workers and senior managers with access to critical upskilling initiatives (should they fall behind in certain areas).
The negative implications are being felt by a different socio-economic cohort. It’s been estimated that over 400,000 jobs were lost to automation in US factories from 1990 to 2007, while another study estimated that 42% of jobs lost due to the pandemic will never come back, primarily low-wage roles often held by minority groups, with those laid off having no time to re-train and or gain access to crucial upskilling.
Nevertheless, as economist Daron Acemoglu notes: “Technological progress is the most important driver of human flourishing but we tend to forget the process is not automatic.” Potential job loss does not mean we should not continue to automate repetitive or low-skilled tasks; merely that it highlights the importance of collaboration between governments, educational institutions, employers, and other organisations to ensure access to upskilling and lifelong learning initiatives for a diverse workforce of various demographics and socio-economic backgrounds.
In terms of what skills are needed, Threadgold noted at the National Leadership Conference that a combination of STEM skills and people skills—such as communication, critical thinking and advanced negotiation skills (as offered by the IMI)—are becoming increasingly valued due to the evolving human-machine partnership. Those who understand how to effectively utilise technology while harnessing their people skills should thrive in this new environment.
However, while these are proactive measures to counter replaceability fears around job loss, dystopian concerns at a deeper, more philosophic level remain—that AI is not going to replace merely our job, but us.
In a separate IMI keynote, futurist Gerd Leonhard noted the eight different types of biological intelligence (emotional, logical, kinaesthetic, etc) that humans exhibit, as compared to AI’s one kind of intelligence (logic). This may be evident in AI’s continuing difficulty to engage in that most basic of things: a spoken conversation. Accents, sarcasm, jokes, metaphors, non-verbal cues, and numerous other hidden complexities that make up a conversation, which we have no difficulty engaging in, cause AI and Machine Learning systems immense difficulty.
And despite the market proliferation of devices like Alexa and Siri, repeated user uptake (beyond one-off novelty use) remains slow. Some of this may be related to usability issues, but there may be something deeper in play.
For example, when Google released a demo of its Duplex virtual assistant in May 2018, it seemed to mark a revolutionary moment in spoken language technology. Combining deep learning, text-to-speak technology and natural language processing, Duplex effortlessly mimicked human speech and casual interaction without needing to rely on previous tricks that successful chatbots had historically relied on, such as relaying the speaker’s words back at them. Yet for this reason, its success at sounding human and engaging in conversation, there was major backlash and Google announced Duplex would announce itself as a bot on future calls.
The backlash to Duplex may be an example of the “Uncanny Valley" effect, whereby a human-seeming artefact triggers feelings of eeriness and repulsion in an observer. As human likeness increases, so affinity increases until a point where artefacts start to become “creepy” and affinity goes negative. Despite numerous explanations offered for this phenomenon, no conclusion has been agreed on. Overlooked may be a misunderstanding of the properties of language and communication.
Language and communication re-inforce our uniqueness and humanity. It’s partly how we express and display our creativity and ability to think critically. We do this through writing as well to an extent, but whereas written communication tends to lean heavily towards message passing and information sharing, spoken language is a much more dynamic and complex beast with so much to figure out beyond the logical.
We don’t notice it because we do it every day, but it takes a huge amount of emotional intelligence and intuition to be able to understand an accent or tell a joke, never mind to intuit body language or the unsaid in a high-pressure negotiation situation. Spoken language interaction and communication between humans is grounded through shared experiences, representations and priors—which highlights the inherent difficulty of constructing a technology intended to replace one of the participants, considering the mismatch between partners.
The “relationship of presence” that’s established during in-person communication—where two or more participants are not only present to each other as subjective beings, with thoughts, feelings, emotions, beliefs, etc. but also engaged in dynamic and reciprocal push-pull influencing (e.g. step towards one partner quickly, the other is likely to step back)—may partly explain the continued demand for in-person learning programmes and networking events held on the IMI campus.
Over Zoom, two participants are still, to an extent, present to each other, but the relationship of presence is lessened through the medium of the digital screen. While digital technologies allow participants to hear and see each other, there can be no reciprocal influencing in the physical sense and any co-presence can be easily ended by exiting the meeting or switching off the laptop.
Getting back to AI, as Gerd Leonhard recently noted, machines cannot feel compassion—at most, they can mimic it (primarily through advances in spoken language technology and Natural Language Processing). Leonhard’s statement will either allay or re-inforce fears around replaceability, but it highlights the importance of developing advanced people skills and communication skills.
Sharing some internal research, IMI has seen a huge uptake in interest for its Advanced Communication Skills programme since the start of the generative AI boom this year. The feeling among participants is that as AI and automation automate more menial tasks, the need to master complex people skills— such as the art of negotiation, to be a more confident communicator, to be more strategic—has never been more important and urgent.
