Everyone’s talking about AI trading bots, how they can enhance fraud detection, optimize liquidity provision, and even analyze market trends in once unfeasible ways. And they’re right, bots can do all these things better than their human operators.

But despite these clear benefits, crypto exchanges should think carefully before proffering proprietary trading bots to retail users. The issue isn’t AI itself but hype-driven bets that trade away explainability and user agency, at real cost to trust, as unchecked ‘AI’ swaps explainability for spectacle with autonomy and trust as collateral.

The in-house development of AI trading bots raises serious concerns about potential conflicts of interest, not to mention heightening regulatory risk. In many regulated markets, a venue cannot provide client-facing automated trading or advice without separate authorization and strict controls.

As such, crypto exchanges should exercise caution before rushing to release proprietary bots just to be seen to be relevant. The tech is great, but sometimes there’s wisdom in holding back.

The Level of AI Bots

AI's infiltration into finance has occurred more quickly than anyone could have imagined. One moment, AI was constrained to a dedicated chat window; the next, it was popping up in our search, our email, and our crypto trading apps.

And to a large extent, our new AI overlords should be welcomed. Tools powered by large language models and neural networks can process vast amounts of data, spot patterns in price charts, and execute trades faster than any human. In crypto, where volatility is a feature, not a bug, this is a big deal.

A bot that can intelligently use technical and sentiment analysis can not only identify the opportune times to enter and exit a trade, but it can also execute them. Because bots, lest we forget, never sleep or grow weary after 12 hours spent switching between screens. They don’t need coffee to run efficiently: just data.

That’s the bull case for bots. But this rosy assessment overlooks several key drawbacks. For one thing, AI bots rely on probabilistic models, which excel at short-term predictions based on historical data but can falter in unpredictable markets. Crypto isn't just about charts! Global events and sentiment influence that no algorithm can fully anticipate with certainty. At least, not yet….

Source.

The Limits of AI Bots

Could a bot determine that your memecoin is dumping because a far-flung nation just bombed a neighbouring country? History shows that automated strategies work until they don't: think of all the quant funds that imploded during market crashes because their models couldn't adapt.

As humans, we can’t easily predict black swan events, but at least we have the humility to acknowledge their potential to strike at any time. If a bot doesn’t see it in its back-tested data, as far as it’s concerned, it’ll never happen. And markets are full of case studies of things that shouldn’t have happened and then did.

But even if we suppose a bot is intelligent enough to surmount all these hurdles and outperform the market no matter what, there’s another reason why exchanges should think twice before rolling out AI trading bots, and this one has nothing to do with the technology itself. This relates to ethics…

A Colossal Conflict of Interest

Pitting AI against AI creates a zero-sum game. If every exchange offers bots with similar capabilities, trading becomes a battle of computing power at best. Those with more resources, institutions, or wealthy users gain an edge, turning crypto back into an elitist arena. For retail investors, it could devolve into gambling: pick a bot and hope for the best. This isn't empowerment: it's a distraction from building genuine financial literacy.

Moreover, if an exchange bot encourages frequent trading to generate fees, whose interests does it serve, the user’s or the platform’s? Would a bot, developed by an exchange, choose to simply buy and hold Bitcoin for a decade —never making a single trade? Offering a bot that trades on behalf of users also blurs the lines between an exchange and an investment advisor, inviting scrutiny. Are we equipped to handle the fiduciary duties that come with that?

Specific tools like grid trading or dollar-cost averaging are fine: they're transparent strategies users can understand and control. But opaque AI bots that decide on your behalf are a step towards managed accounts, which many exchanges simply aren't licensed for.

Kucoin is one of many exchanges to offer users access to trading bots.

Source.

Empowering Users, Not Replacing Them

At its core, crypto is about empowerment and self-sovereignty. Exchanges should enable that by giving users optionality, not by automating decisions away. This approach keeps control in the hands of users, fostering skills like risk management and strategy development.

AI undoubtedly has a role in efficiency: automating tedious tasks like compliance checks or portfolio rebalancing frees up time for smarter decisions. But over-relying on bots could erode that judgment, which is critical in a field as dynamic as crypto. Developing personal capacities, like understanding market fundamentals, is far more rewarding than outsourcing everything to code.

Lessons From Blockchain's Hype Cycle

Remember the pre-ICO era, when every company was grilled on its “blockchain strategy”? Many shoehorned it in unnecessarily, and if not reined in, AI risks a similar fate in crypto. Just because the technology exists doesn't mean exchanges should be sticking it to everything like a band-aid. Does it solve a real problem, or is it just buzzword-chasing?

AI and crypto can complement one another powerfully, from better analytics to streamlined operations. But when it comes to trading bots, exchanges should tread carefully to avoid conflicts and diminished user agency. That’s why their focus should remain on building tools that empower rather than extract.

Let's embrace AI thoughtfully, ensuring it serves the community rather than the crypto exchanges that have commissioned it. Generating short-term hype will carry you so far. But building trust will carry you a long way.

This article was published under HackerNoon’s business blogging program.