AI won’t take your job. But a human using AI will.
It’s quickly becoming a cliche in the AI-jobs debate.
But it ignores the type of human and the type of job. Take sales. What makes someone good at it? Emotional intelligence, curiosity, resilience and amiability. Even if your best salesperson eschews AI, you’re unlikely to find a better replacement in an AI devotee without these skills. She may be able to scramble out 1,000s more cold emails and track leads more effectively. But quantity and diligence doesn’t displace relationships.
Baillie Gifford’s Tom S. talks about jobs as a bundle of tasks. On Merryn Somerset Webb ’s Bloomberg podcast, he advises individuals and companies to think about which of these tasks are better done by AI, and which by humans.
Excelling at the latter means setting some boundaries with AI. Outsource your thinking too often and you lose the ability to think at all. Tom gives the examples of sat navs, which all drivers now rely on. Tom can no longer rely on memory for certain routes because regular sat nav use has wasted away this ability.
Maybe no great loss in a quantitative capacity. If your priority is getting from A to B as quickly as possible, Google Maps is a better bet than experience. But what does it mean for qualitative decisions? For instance, I regularly see professionals using AI to clear their inboxes faster. Copying and pasting inbound and outbound emails to give the veneer of professionalism. The result is often to amplify meaningless chains, rather than exercise judgement on any course of action.
Tom refers to MIT and Lancet studies that both show the pernicious effects of such outsourcing. In the MIT study, participants were tasked with writing essays with and without AI. Over 80% of the AI-assisted group couldn’t provide a single quote from essays they wrote minutes earlier. In the Lancet study, endoscopists’ diagnoses were 21% less accurate with AI. Performance collapsed as they unquestionably trusted confident, but inaccurate, interpretations.
It’s a consequence of the Dunning-Kruger effect. The more you know, the more you know the extent of what you don’t know. So real experts in a field grapple with uncertainty whereas novices lack the self-awareness of their own ignorance. They’re far more likely to blithely take any LLM response at face value.
Hence, AI isn’t a rising tide that lifts all boats. It’s a force multiplier. These tools are fantastic when you know what you’re trying to do and can interrogate responses. So yes, our talented salesperson benefits when AI uplifts organisation skills (not that they’re all chaotic…). But equally, she’ll exercise her own judgement on when and how to contact prospects. And that’ll stand her in rather better stead than the faceless replacement sending out an army of agents to harass the CRM system.
What are the lessons for individuals and corporations today?
- • Individuals: Don’t get compressed to the median. If everyone is using AI, think about competitive advantages. Namely originality, novelty and deep expertise. Learning and curiosity are more important than ever.
- • Corporations: Be wary of chasing short-term rewards. Cutting the workforce plays very well in the market right now. But what are the long-term consequences of failing to invest in human capital?
Prompt engineering has its place. AI natives are pushing the boundaries of what’s possible with these tools. But that’s enabled by a worldliness and knowledge that enables such actors to know what they want to create. Engaging better with AI means bettering yourself too.
AI won’t take your job. But a better human using AI will.
