The Digital Toolbox: Why White-Collar Workers Are Becoming Tradespeople

Tradespeople have always carried their own toolbox. Power tools, socket sets, PPE – the things required to complete the vast majority of their tasks. These tools travel with the worker from job to job, and provide a source of pride, competitive advantage, and a barrier to entry against newcomers. Those fancy Snap-On tools which fixed your Prado were probably owned by the mechanic, not the shop.

I’m told this is not an acceptable tax deduction for someone in my line of work.

But for us white collar workers at the PowerPoint factory? We just showed up. A company laptop was provided with company subscriptions and a company database. The distinction between company resources and personal resources was clear, which is handy as it’s important in determining who owns what and what our relationship is.

But the old order no longer applies. Many modern white-collar workers bring with them a digital toolkit of subscriptions, memberships, and capabilities to complete their work, above and beyond the resources provided to them by the company.

The key tool in this kit is, of course, AI. 20% of AI usage at work is with personal accounts, not accounts provided by their employer. Around half of Australian workers admit to using AI in a way which contravenes company policy.

It’s running emails through ChatGPT, getting Perplexity to explain a topic, using NotebookLM to summarise a big stack of papers. You’ve seen it, you’ve probably done it. It’s hard to avoid and it’s hard to resist.

Employees may not disclose to their employers the extent of their personal digital toolkit– why would they? The employee gets the benefit of looking more efficient, and the employer gets the benefit of that efficiency. I suggest that this has quietly become the norm, and the consequences have not been fully considered.

Employers must know that their employees are using their own AI tools on the side, but I think they would rather not say anything explicit about it while the productivity keeps rolling in.

Bask in the efficiency.

There are all sorts of sticky questions involved in the usage of AI which most companies don’t have the time for. And, if nothing else, it might be handy to know that your employee has been breaching a code of conduct, just in case they’re no longer required down the line.

The idea that one can outsource their work entirely isn’t a new concept – paying people in the developing world to do your job for you has been around at least since The Four-Hour Workweek. The difference here is the scale. It was always a cheeky, clever-but-weird move pulled by that one guy you knew. Now, it’s Keith in accounts receivable.

This normalization of the ‘personal digital toolkit’ raises a few questions which might be of interest to legal ponderers. Does using one’s own AI subscription impact who owns the IP in an innovation made at work? What about the employee vs contractor characterisation? Where exactly is the line on the transmission of confidential client data to an LLM – if key information is anonymized, does that make it OK? And, perhaps most importantly, how many decision-makers are thinking about these things proactively?

More broadly, I see it part of a larger trend of white-collar employees being pushed more towards the role of an itinerant tradesperson. I am just old enough to have been around when the work smartphone was introduced. The idea was that it would increase our flexibility, but the unwritten expectation was that individual employees would now be responsible for managing their own working hours, much as a contractor does.

Things have only progressed from there, and the next thing to be pushed onto white collar employees may well be the responsibility for managing their own intellectual toolkit. I think that the unwritten expectation might become that all workers bring with them their own little taskforce of secret LLMs. Such expectations can be easily buried beneath vague statements about ‘productivity’ and being ‘technology adept’. After all, if you’re not paying for and using your own LLMs, maybe your replacement will. And Keith is looking at a new jetski.

It’s tricky to work out the appropriate analogy to apply here. Maybe it’s as minor as a worker bringing their own calculator, or as major as being pushed by into drug use to keep up the pace. To me, the best analogy feels like the worker using a precedent from an old job, a pirated version of a program work won’t pay for, or a contact list pilfered on the way out the door of a previous job. The work’s getting done, even if the means are a bit questionable. Nobody’s saying it out loud, but everyone’s benefitting, and everyone’s complicit.

This is currently a question of personal responsibility for individual workers. The incentives are pretty clearly aligned towards the quiet use of personal AI systems in the workplace, for each employee to have their own personal digital toolkit. The circuit-breaker would be the widespread rollout of enterprise AI solutions which can give the same or better outcomes for all parties. No wonder the legacy corporations are spending so much on it.

We’re still in the wild west for now, though, and there will always be an incentive for individual employees to jump on the next wave to improve their own output in a tight labour market. So the wild west might be around for a while.

We’ll get to the end of it someday.

Employers can probably pretend that the personal digital toolkit isn’t happening, but is there an equivalence between looking the other way on workplace safety, and a lax attitude towards AI? Even when mechanics use their own tools, employers still have a duty to provide a safe working environment, and a responsibility to clients to deliver acceptable service. Surely the same standards apply?

An employer can very well say that they didn’t know what was going on when a worker uses their personal digital toolkit in an objectionable way – but how long before nobody believes the employer, when everyone knows what’s happening across the workforce?

Or, to put it in a very ChatGPT kind of way: It’s not a question of whether employees are using their own personal toolkits – it’s a question of whether employers will take that responsibility before they’re forced to do so by a judge.