Why AI Does Not Automatically Make Legal Services Cheaper

“AI does not solve the problems of the legal sector. It exposes and amplifies them.”

Many predictions about AI in the legal sector sound promising: cheaper services, better access to justice, and faster procedures. But is that actually true? According to a new research report by Lawfare, written by researchers from Harvard Law School and Princeton University, the answer is more nuanced than many hope. Advanced AI certainly makes legal work faster and cheaper, but that does not automatically translate into better or more affordable outcomes for the client. The researchers identify three structural bottlenecks that stand between technological progress and actual improvement of legal services.

Why Legal Services Are So Expensive

Before we look at AI, it is important to understand why legal services are structurally expensive.

The report points to three causes:

• Legal services are difficult to evaluate. Unlike a product in a store, a client often does not know whether legal work has been performed properly. Even in retrospect, this is difficult to determine. Clients therefore rely on reputation and prestige when choosing a lawyer, not on measurable quality.
• The value of legal work is relative. In a dispute, it is not about how good your lawyer is, but whether they are better than the opposing party’s lawyer. This drives an arms race: both parties invest increasingly more to maintain an advantage.
• Regulation limits competition. Frameworks regarding scope of practice within the legal profession and ownership restrictions for law firms hinder new entrants and alternative business models. This keeps prices artificially high.

The report illustrates this with a striking figure from the American market: average partner rates at large firms are now above $2,300 per hour. And even at the lower end of the market, lawyers spend only 2.3 hours of their workday on billable work, while clients pay the full rate for this.

Three Bottlenecks Between AI and Affordable Justice

The report’s central argument is clear: even as AI becomes increasingly proficient at legal work, three bottlenecks stand in the way that prevent this progress from reaching the client.

1. Regulatory Barriers

In the United States, rules regarding the unauthorized practice of law stipulate that only qualified lawyers may perform legal work. This also affects AI applications: as soon as an AI tool performs tasks that can be classified as ‘legal advice,’ the provider risks fines or criminal prosecution. The report cites the example of LegalZoom, which faced years of litigation because it automated simple legal documents. Non-profit organizations that deploy volunteers to help citizens with debt collection cases also encountered legal restrictions. The result: innovation is discouraged and consumers who would benefit most from cheaper legal assistance are denied access to it.

2. The Dynamics of Procedural Law

The second and perhaps most surprising bottleneck is the arms race that arises when both parties in a dispute deploy AI. In a system where the value of legal work is relative, greater productivity does not automatically mean a better outcome.

The report makes an essential distinction:

• Production is what legal work yields: contracts, pleadings, and briefs.
• Outcomes are what clients truly want: a resolved dispute, a closed deal, protected rights.

When both parties deploy AI to produce more and faster legal work, the equilibrium simply shifts to a higher level. The result remains the same, but the amount of work required to achieve that result increases.

The researchers draw a historical parallel with the digitization of the discovery process in the 1990s. This could have reduced the costs of evidence gathering, as documents became searchable more quickly. In practice, the opposite happened: parties used the explosion of digital documents precisely to overwhelm the opposing party with information, causing total costs to rise.

This pattern threatens to repeat itself with AI. If it becomes cheaper to produce legal work, this could lead to:

• Longer and more complex contracts (between 1996 and 2016, the size of acquisition agreements in the US grew from 35 to 88 pages).
• More pleadings and motions in disputes.
• A further escalation of the already existing arms race for legal talent.

The hourly rate reinforces this dynamic: more hours mean more revenue for the firm, without the client necessarily getting a better result.

3. The Necessity of Human Oversight

The third bottleneck concerns the speed at which humans make decisions.

Even if AI performs all legal tasks lightning-fast and flawlessly, a human limitation remains:

• Judges need time to assess cases. If AI causes a flood of new proceedings, courts become overloaded. The consequence: longer processing times or lower quality of adjudication.
• Lawyers and clients must understand complex contracts. If AI can draft fifty clauses in an instant, a human must still read, understand, and evaluate them.

The researchers estimate that AI could lead to a doubling to quintupling of the number of judicial proceedings. This places the judiciary under enormous pressure and raises the question: do we want machines to take over the administration of justice? The report is clear on this: human involvement in the legal process is valuable and necessary.

Three Considerations for Legal Professionals

1. Think in outcomes, not production. AI makes it tempting to produce more legal work. But the question must always be: does this lead to a better result for the client? Focus the deployment of AI on outcomes, not output.
2. Anticipate the arms race. If your firm deploys AI, the opposing party likely will as well. Consider how AI changes the dynamics of disputes and negotiations and adapt your strategy accordingly.
3. Invest in human judgment. Precisely because AI is increasingly taking over executive tasks, the human factor is becoming more valuable. Invest in strategic thinking, complex decision-making, and the ability to critically assess AI results.

AI Is Not a Miracle Cure, But a Magnifying Glass

This report offers a welcome nuance in a debate often dominated by optimism. AI has enormous potential for the legal sector, but that potential is only realized if the sector also addresses structural problems. Those who think that AI automatically leads to cheaper and better legal services underestimate the power of existing structures.

The future of the legal sector is not determined by the technology itself, but by how the sector deploys that technology. This requires leadership, reforms, and a willingness to break existing patterns. Not only in the US, but also here in the Netherlands.

AI is not a miracle cure. It is a magnifying glass: it makes visible what has been askew for some time. The question is not whether AI makes legal work cheaper. The question is whether we are willing to implement the changes necessary to make that work for the client as well.