Episode 1 - But AI Said It Would Work...

Over the past year, one phrase has become increasingly familiar in our conversations with Hi-Fi customers:

“But AI said it would work…”

We’re seeing more customers arrive with AI-generated recommendations, sometimes changing direction several times before they even make a purchase. And occasionally, we’re helping people understand why equipment that seemed compatible on paper has disappointed them in practice.

To be clear: we love AI. We use it ourselves, and we think you should too. It is brilliant for gathering information, understanding terminology and narrowing down options. The problem begins when useful research is mistaken for a final, independently verified verdict.

We use AI ourselves. Here’s a perfect example.

Recently we needed a wiring loom for a car head unit. We gave AI the make, model, year and relevant details, and asked it to help find the correct loom. It was excellent for identifying possibilities and saving research time. We could then verify the fitment before ordering.

But asking “Which wiring loom fits?” is a different kind of question from “Which head unit sounds the best?” One is largely a compatibility search. The other involves performance, system integration and personal judgement. AI can contribute to both — but the second requires more scrutiny.

Use AI to do the heavy lifting. Then verify the result.

A recommendation can make perfect sense — and still be wrong.

Take a familiar amplifier-and-speaker matching problem. Imagine a loudspeaker listed at 89dB sensitivity and 4Ω nominal impedance. A customer asks whether their amplifier can drive it. Based on the published figures and an amplifier rated for 4Ω operation, the answer might sound straightforward.

But what if the loudspeaker’s impedance dips towards 2Ω at certain frequencies? What if the electrical phase angle makes the load still more demanding? And what if the amplifier lacks the current capability or headroom needed for that particular combination at the customer’s preferred listening level?

Those details may not appear in a basic specification listing. Moreover, sensitivity and nominal impedance are different measurements: an 89dB rating does not establish that a speaker is easy to drive.

A customer may then describe the system as lacking weight, authority, depth or dynamics. Under some conditions an amplifier facing a difficult load can compress or distort, but those subjective symptoms can also come from room acoustics, positioning, recordings or setup.

That is exactly why proper diagnosis matters. If the amplifier is the issue, buying another pair of loudspeakers could be an expensive way of avoiding the real problem. We’ve encountered customers struggling with unexpected system matching; the figures above illustrate the principle rather than reproducing one named customer’s measurements.

But who is AI actually taking advice from?

AI can draw on technical data, professional reviews, forum discussions and patterns learned from a large amount of published material. Some of those sources are exceptionally valuable. Others deserve caution.

Imagine a forum contributor confidently writes: “Amplifier X sounds incredible with Speaker Y.” Perhaps they have owned the combination for years, carefully compared alternatives and know exactly what they are hearing. Or perhaps they have never heard either product and are repeating something somebody else said.

Without evidence, we cannot tell. Nor can an AI assistant reliably establish every contributor’s qualifications, motivation or firsthand experience.

When an opinion is repeated across multiple places, it may begin to look like a consensus. But ten websites repeating the same original claim are not ten independent listening tests. The volume of online information is not the same thing as its quality.

The most visible product is not necessarily the best match.

There is another question worth asking: could a manufacturer with more reviews, interviews, videos and searchable product information be easier for an AI system to discover than a smaller specialist brand? Certainly, online visibility can affect what information is available to evaluate.

That does not mean companies can simply purchase favourable AI answers, nor that AI always recommends the most advertised product. But the online information landscape is uneven. Some exceptional equipment receives very little publicity; other equipment generates enormous amounts of coverage.

At Kronos AV, we work with well-known international names and smaller specialist manufacturers. Sometimes the most suitable option is also the most famous. Sometimes it isn’t. Popularity and system suitability are two different things.

Why “But AI said…” can become an expensive cycle

One common pattern is that a customer receives an AI recommendation, asks about another option and gets an equally persuasive explanation, then asks about a third. Each answer may be reasonable under a different set of assumptions, but the conversation can start to feel like a constantly changing final verdict.

The customer can become stuck in an endless loop of alternatives. If a purchase disappoints, they may seek another recommendation and replace a component before understanding what actually went wrong.

More recommendations do not necessarily mean a better decision. At some point, you need to test the assumptions and check the real-world evidence.

The most important question: is it right for YOU?

There’s something we’re always careful to explain in our own Kronos AV Review Diaries: our opinions come from genuine listening and practical experience, but every customer’s circumstances are different. An amplifier we love, or a speaker we rate highly, won’t automatically be the right choice for everybody.

Your listening room, your existing system, the way you listen, your budget and even your home network can change the answer. Sometimes the best recommendation isn't just “yes, it works.” It's “yes, provided you have the right conditions in place.”

Take a music streamer. It may be advertised as wireless, support high-resolution playback and appear ideal on a specification sheet. But perhaps our experience with that model suggests it is more reliable with a wired connection, or a particular streaming feature has a limitation depending on the app or connection used. We might happily recommend it assuming the customer has suitable Wi-Fi coverage, a reliable network or an Ethernet connection available. Without that, we’d want to discuss another approach.

That doesn’t mean wireless streaming is inherently poor, or that everyone needs a mesh network. It means the details of the customer’s setup matter.

AI can ask excellent follow-up questions and make increasingly personalised suggestions. But customers don't always know what information to volunteer in the first place. An experienced dealer can sit down, work through the complete system, ask about those easily overlooked details and explain not just whether something is compatible, but why it is — or isn't — right for that particular person.

We welcome advice from reviews, forums, AI and other experts. We offer our own opinions too. But none of those opinions is more important than the outcome for the person who will actually be living with the system.

So should you stop using AI for Hi-Fi?

Absolutely not. Research your options. Compare features. Explore unfamiliar manufacturers. Learn what terminology means. Build a shortlist, and use AI to ask better questions.

Then, before spending £2,000, £5,000 or £10,000, take two minutes to verify the recommendation with somebody who works with the equipment.

“AI recommended this amplifier with my speakers. Do you think it’s a good match?”

We might say yes — and explain why. We might flag a possible compatibility issue or suggest a different option worth hearing. If we haven’t tested that exact combination, we’ll say so. The point is not to replace one unquestioned opinion with another: it’s to help you make an informed choice.

This reflects one of the core ideas behind Kronos AV: helping people avoid expensive mistakes. Our job is not simply to sell you an amplifier or a loudspeaker. It is to help you end up with a system you enjoy, without having to buy everything twice.