Op-Ed: AI in consumer dispute resolution? Unexpected upsides

AI in consumer dispute resolution generates some instant natural responses. Trust in AI is low enough. But what if you create a regulated dispute management system from scratch, with compliance built in?

Disputes are common enough. They’re time-consuming, frustrating, and expensive in time and sanity. There’s a lot to be said for a neutral arbiter that can streamline the process. Hence the somewhat pragmatic surge of interest in AI as an option for dispute resolution.

The simple fact is that everyone prefers things to run on rails and be clearly rules-based. Nobody wants or needs a hyper-legalistic disputes system where nothing is actually resolved.

It’s slowly and haphazardly happening, but it’s happening. That’s the general framework of an embryonic start-from-scratch idea now fermenting in Australia based on modifying current dispute resolution rules in various sectors.

In Europe, the Digital Services Act (DSA) has now been finding its way with mixed results, but out-of-court dispute settlements have obvious benefits for all parties.

British Columbia’s Civil Resolution Tribunal guides people through an interactive Solution Explorer, a surprisingly broad-based “rules based chatbot”.

Enquiries about AI dispute resolution get a pretty blunt response from various AIs, which all cite significant issues with decision-making.

Gemini spells out benefits with clear caveats.

Copilot is more analytical about the dynamics of AI dispute resolution and realistic about the limitations of AI managing emotionally charged conflicts.

Grok goes quite a bit further in delineating challenges, risks, and limitations.

They agree that AI can streamline the process and manage the range of disparate data that disputes inevitably create. Bias and transparency are the main issues.

Grok points out that in terms of accountability and final decision-making, “Most systems keep humans responsible for outcomes. Fully autonomous “AI judges” remain rare and controversial.“

That’s a fair summary. In Europe, the AI can’t be the sole decision-maker. It is unrealistic. Dispute is the basis of civil law, and humans have been working on it for thousands of years with mixed results.

AI can negotiate disputes? Yes, but…

In some cases, AI can act in an advisory capacity for negotiation of disputes. There’s a cost-benefit in time and speed of resolution, but only if the negotiations succeed. The contrast with human mediation is stark.

If you’re trained in negotiation, you may see differing values and risks in the whole theory of AI negotiating disputes:

Negotiation has to show clear benefits to both sides to work at all.

Negotiation is the diametric opposite of a simple rules-based decision, which may or may not be acceptable to parties in a dispute.

Participation in negotiation can turn itself off, leading nowhere but to an arbitrary decision.

The bargaining process in any dispute can be very tricky and value-based, particularly when assets are involved.

Less obvious is that this level of flexibility in AI responses may open up ways to settlement outside the pure complaint resolution framework. Sometimes the rules don’t really match the circumstances or the preferences of parties.

Sometimes humans get stuck in immovable Yes/No situations where neither party can agree with the other’s proposals, but they still want resolution. This “AI fourth party” approach can do that.

There’s an irony here. When describing AI dispute resolution, all the Ais noted that there was a risk of bias in AI processes.  In this scenario, a neutral settlement is more likely to work than a one-side-or-the-other approach.

Parties can accept that an impartial AI isn’t acting on its own platform’s behalf. It’s a vindication of sorts for AI dispute resolution, at a distance.

Rules-based AI has a point to make

In the compliance arena, AI does have a lot to offer. The main unresolved issue is AI compliance with the vast range of laws worldwide. At the local level, however, compliance is simpler and easier to define.

This is what rules-based AI can do:

Indexing and citing relevant laws, regulations, and rules: This is a very tedious process for humans, and not always accurate. AI can do most of that work in seconds.

Establishing patterns of complaint: This is an invaluable resource for organizations to pin down emerging issues and define the volume of disputes.

Enabling broad-based problem-solving: If there’s a series of complaints and disputes, the AI can deliver statistics, outcomes of dispute resolutions as they arise, and define success or failure of resolution attempts.

These are number-crunching processes, well suited to AI processing capacity. It’s a type of market research, too, evaluating issues on a clear statistical basis.

For example, AI can generate a clear picture of problems almost instantly and on an ongoing basis:

Widget A generates 100 complaints within an hour of release onto the market, scaling up to 2000 in 5 hours.

Complaints can’t be managed in-house at sales points because it turns out the widgets need technical tweaks.

Resellers are getting deluged by complaints in 24 hours.

Individual complaint resolution could take months on this scale.

The manufacturer and distributors need a clear, simple fix.

Solution: Refunds or offer to replace Widget A with tweaked versions. This is easy to do with any level of chatbots, and can be managed at all degrees of difficulty with cost data built in.

In this case, it’s the ability of AI to manage factual data that allows AI to handle the disputes quickly and efficiently within statutory rules.

That’s the real issue with AI dispute resolution. Never lose sight of the fact that AI, like people, works better in a rules-based system.

.