‘I don’t buy it... it's AI-washing’

Block’s CEO Jack Dorsey's plans to cut nearly half of his 10,000-person workforce, while talking up artificial intelligence and posting profits, seems to preview where white-collar work could be headed.  

In a memo posted on X, Dorsey said he would cut Block’s workforce to just under 6,000, even though “the business [is] strong and profits growing,” arguing that one large reduction is better for morale than repeated rounds of layoffs.  

For Samuel Dahan, associate professor at Queen’s University Faculty of Law and director of the Conflict Analytics Lab, the AI rationale rings hollow when set against a broader economic backdrop.  

Dahan has spent the last decade building and studying legal AI applications, and he sees a clear disconnect between what these systems can do and how they are now being used as an explanation for job cuts.  

AI-washing and the real drivers of layoffs 

“I don’t buy it,” Dahan says of the layoffs, pointing instead to economic pressures that predate the latest wave of generative tools. 

“I think a lot of it is AI-washing, and they’re essentially attributing layoffs to AI that might have happened regardless.” 

In his view, Block’s announcement lands in a broader “cost-cutting cycle” that echoes 2019, 2008 and the post-dot-com era, with high interest rates, war, tariffs, supply chain volatility and “investor pressuring for margin discipline” all in the mix.  

Business Insider reported that Block’s shares jumped more than 20 percent following the announcement, with a Stanford finance professor suggesting a CEO “race” to please investors. 

Economic forces, Dahan says, would have prompted some level of retrenchment with or without AI. He acknowledges that AI has become part of the story – just not in the way many corporate statements suggest: “It’s not because of what AI can do yet. It’s because of their belief – well actually, what they want to believe, or what they want the investors to believe.” 

Evidence, not belief, should drive AI workforce calls 

Business Insider reported the internal story Dorsey told employees and shareholders; in his memo, he acknowledged that repeated rounds of cuts are “destructive to morale” and says he prefers “a hard, clear action now” over “a slow reduction of people toward the same outcome”.  

For Dahan, that juxtaposition – AI as public justification, cost-cutting and investor reassurance as the underlying logic – should tell Canadian employers to separate rhetoric from reality before attributing workforce changes to technology. 

Resist making headcount decisions primarily to satisfy that investor race or because peers are doing the same, he says; the current state of AI adoption does not justify treating large groups of roles as obsolete.  

“Right now, the real issue is that we don’t really have real evidence that it’s truly adding productivity,” Dahan says. 

“Make decisions based on a data-driven analysis. Don’t make decisions based on fear or FOMO or what everybody else is doing … because a lot of studies are showing very mixed evidence in terms of productivity, mixed evidence in terms of ROI, mixed evidence in terms of implementation of the pilots that have been tried over the last few years.” 

For employers in Canada, he says, that means building internal measures of AI impact on specific tasks and teams, rather than assuming global headlines about productivity gains automatically apply to their own workforce. 

Rethinking jobs as tasks, not headcount 

Behind the headlines, Dahan says the real transformation is at the level of work design, not wholesale job elimination. Rather than viewing a role as a single unit to be “replaced” by AI, he explains that technology tends to affect specific activities within that role: “Think about jobs in terms of tasks, not in terms of job as a whole.” 

That framing should help employers see opportunities to reallocate work instead of defaulting to layoffs. Even where AI tools are getting better at drafting emails or generating first-draft reports, he emphasises that they still require people in the loop.  

“You still need a lot of people to orchestrate these models,” he says.  

“We’re not at a stage where you can have agents doing pretty much everything on their own. Just going to your email, writing the email, connecting with Slack and connecting. They can do that, but they need a bit of orchestration, the full automation is not happening yet.” 

That shift shows up clearly in his own field. In legal work, he says, a new evaluative role is very different from traditional drafting and research tasks, and calls for updated training and supervision models. For junior people, this is a significant adjustment that Canadian employers will need to plan for, he adds, because appropriate AI literacy requires experience: “They’re not really equipped to evaluate AI output as much as a more senior person.” 

What Canadian HR should do before reducing headcount 

Dorsey's memo presents AI as a maturing capability that justifies a much leaner organisation. In contrast, Dahan’s experience in legal AI suggests most companies are still at the beginning of the journey – and that their buying decisions can easily outpace their actual needs.  

Many vendors now pitch specialised “AI for HR” products that promise quick results, but Dahan believes employers should look carefully at what is truly under the hood. To him, the most strategic question for HR and IT is not which branded tool to pick, but how they will retain control over their data, workflows and governance as they build on top of general-purpose systems.  

“Technology doesn’t fix a broken system,” Dahan says; his core message for employers is to focus on systems and evidence long before they consider AI-related layoffs, especially hard-hitting ones. 

“If the system is broken, adding a layer of technology – it could be AI, it was Zoom during the pandemic, before that it was data science – I think it’s probably a deeper problem,” he says. 

“Just make a decision based on evidence, not based on beliefs.”