Strategies which rest on the assumption that AI simply makes everything faster, cheaper, and easier are ones that risk being dangerously simplistic and overlook one inconvenient truth: AI reshapes work at least as much as it streamlines it.
For in-house teams allocating budget and assessing risk, the following counter-intuitive truths demand attention.
Your heaviest AI users may be working more, not less
The promise is seductive: deploy AI, encourage its use and watch working hours drop. Alas, the reality is often the opposite. Take super-users:
These are users who are especially curious about AI and experiment extensively with it. They have no formal qualifications in AI, but they see opportunities to integrate AI into their daily workflows and proactively look for new ones.
Research from 2026 by Aruna Ranganathan and Xingqi Maggie Ye, AI Doesn't Reduce Work - It Intensifies It, reported in the Harvard Business Review, showed that super-users can end up logging more hours than their peers. They do so voluntarily, apparently because the work becomes more engaging when the drudgery falls away.
In-house teams should track how AI adoption affects actual working hours, and not just output, and ensure that productivity gains are translating into sustainable working patterns rather than simply more work.
The Solow Paradox has arrived in legal
Economist Robert Solow once quipped that "you can see the computer age everywhere but in the productivity statistics."
AI now inhabits the same paradox – for now.
Self-reported gains run far ahead of measurable ones. As always, in-house teams should treat vendor productivity claims with a degree of scepticism and demand peer-reviewed evidence before restructuring business models on the back of them.
AI generates a flood of new work
A greater ability to process documents with genAI means more drafting, more markup, more emails, more Teams calls with AI-generated action lists, and, as Lord Briggs warned in April 2026, a likely "tsunami of small to moderate value civil claims" once litigants can formulate them without lawyers.
On 6 August 2026, The Economist noted that citizens are using AI "to file objections and appeals, and claim their dues" from authorities. The result is that the state is drowning in AI-generated slop and could be "brought to a halt". Tarun Tawakley, Partner & Co-Head of Technology at Lewis Silkin, is quoted in this piece on new employment legislation, noting that, "every factor points to more litigation, and more complex and high-value litigation."
If this sounds familiar, remember how email was heralded as streamlining communication in the 90s, but the marginal cost of sending emails resulted in an explosion of them.
Put simply, AI doesn't just accelerate existing tasks; it manufactures new ones. In-house teams should be alive to, and prepare for, rising volumes of correspondence, counterparty mark-ups, and so on.
And don't forget, above all, that AI itself can also be used to help you deal with increased AI-powered workloads – a case of fighting fire with fire!
Intelligence is now a commodity
Our whole society, and the legal profession, is built on the premise that humans have the monopoly on intelligence. This assumption is now obsolete.
If intelligence is no longer scarce, then what is scarce – judgment, relationships, creativity, ethical reasoning – must anchor every in-house team's value proposition.
Don't forget, when developing training plans, to focus on these ever-more-essential skills.
This is a change-management problem, not a technology problem
Lewis Silkin in our Future @ Work 2026 report, drawn from a quantitative survey of 663 senior leaders from the UK, the US, APAC, and the EU) noted that almost half of organisations (49 per cent):
Expect cultural resistance to slow down the adoption of new technologies and the implementation of new operating models.
The pattern is familiar from every prior technology wave: the tools work, but the humans around them often struggle to adapt.
The result? Governance frameworks, incentive structures, protected training time, and visible leadership commitment matter more than the sophistication of the AI model that your team is using or looking to use. If any of these are missing, prioritise them ahead of your next technology purchase.
Too much AI can cause "brain fry"
There is a sweet spot with AI use: excessive task-switching between human and AI workflows can trigger what academics call cognitive overload or "brain fry". Firms must monitor this carefully and consider their approaches to rest and recovery.
The implication for employers is serious: an always-on, AI-augmented working environment may generate wellbeing and performance risks that aren't mentioned in any AI vendor's marketing materials.
The real danger is not hallucination, it's misgrounding
Hallucinated cases (eg, entirely fabricated citations) grab headlines, but misgrounding is the subtler and more dangerous failure mode. This is where AI cites a case that genuinely exists, but which doesn't actually support the proposition for which it is cited. Such errors can slip past verification checks designed to confirm a case is real without testing whether it is relevant.
The result? Supervision protocols must go beyond "does this case exist?" to "does it say what we claim it says?"
Lawyers' habits may be holding them back
Lawyers often default to treating AI as a keen but junior trainee: give it a task, review the output, and move on.
But we should arguably demand far more, such as requesting step-by-step reasoning plans, or ten alternative clause drafts, or a structured critique of our own work. Equally, the instinct to type rather than speak, or to delegate in familiar patterns, often prevents us from discovering workflows where AI delivers genuine value.
Training must tackle ingrained professional habits, not just technical competence. As tempting as it is just to use AI to refine how things are already done, now might also be the time to rip up our internal rulebooks and rethink how we do what we do.
Successful AI adoption requires investment in human wellbeing
AI will, in theory, free up time for more strategic thinking. However, humans are not robots. It is often in our downtime, when our minds wander, that we come up with our best ideas.
This is perhaps the most counter-intuitive finding of all. In the race to adopt AI, in-house teams must simultaneously invest in brains, bodies, and mental health.
Why? Humans often generate their best ideas during downtime, not while processing AI outputs at pace. With 59 per cent of legal professionals reporting poor mental wellbeing in 2025, the risk is that AI-intensified workflows compound an existing crisis.
The upshot? Treat any sustained decline in wellbeing to recalibrate the pace of adoption.
AI training never ends
Today's AI model is, in one of Silicon Valley's favourite phrases, "the worst you will ever use".
Some skills that matter in 2026 may be outdated by 2027. Firms that treat AI training as a one-off onboarding exercise (ie a tick-box followed by business as usual) will find their people falling steadily behind. Continuous investment, protected time, and regular reassessment are non-negotiable.
But sustaining momentum brings its own challenges. There is a real risk of AI fatigue: teams switch off when they feel they're hearing the same message on repeat. The answer is variety and relevance: rotate formats, spotlight new wins, and tie each learning initiative to a concrete business problem rather than abstract capability.
Crucially, teams should be encouraged to revisit use cases that disappointed on first attempt. AI models improve rapidly, and a workflow that produced unreliable results six months ago may now deliver genuine value.
~~~
The collective message might be summed up in a single, uncomfortable truth: AI makes everything different, not simply better.
AI doesn't simply slot into existing structures and deliver savings. It reshapes work patterns, intensifies cognitive demands, creates new categories of risk, and requires perpetual organisational adaptation. In-house teams should calibrate their expectations accordingly.
The firms and in-house teams that will thrive are those willing to treat AI adoption as an ongoing discipline – part technology, part psychology, part change management – rather than a procurement decision with a fixed implementation date. As always, it's about people and AI, never people or AI.
We look at these key issues in much more detail in our book Rising Roles in Law: Designing the AI-Enabled, Data-Driven Law Firm (Globe Law and Business, 2026). For more details, check it out here.
In our next article, we look at these challenges through the lens of legal ops.
If you have any questions or thoughts on this article, please do get in touch. We'd love to hear your experiences.
