Things you bought based on AI advice (good or bad!)

[YouDontHaveEnoughBass]

fancy autocomplete

I think there comes a time where viewing things a certain way is no longer useful

For example room acoustics; every frequency is modal but there comes a point (usually around 150hz or so) where the modal behaviour is so incredibly complex it’s no longer useful to think about it in those terms so we don’t bother

I think we’ve reached this point with AI; yes it can still technically be described as auto complete but its capabilities have grown so complex it’s not a useful descriptor any more

 

[Curvature]

I think perhaps some articles or other media while doing research.

 

[Superparati]

Everyone uses AI to ask questions and/or create images etc, but I wonder if others like me used AI to research a topic and ended up buying something based on its advice? Can be audio related or other stuff. And if it was a good advice or bad? Any AI: ChatGPT, Gemini, Claude, Grok, whatever.

I start with 2 examples that I "discovered" based on AI.

minidsp Flex HTx was not on my radar at all when I was researching for an AVP to replace my old AVR in my HT setup. Of course I know the brand and its most famous products, but particularly the Flex line flew under my radar for some reason. After chatting back and forth with AI for my use case, preferences, budget, limitations, etc I ended up buying the Flex HTx and it worked like a charm (given that you pair it with an Apple TV 4k, which I knew and did). This one worked alright.

The others are not audio related.

I have an old mini ITX computer that I wanted to upgrade its CPU to the maximum the motherboard would accept for light gaming for the kids. It suggested me a specific model (5775C) which I would never guess or have researched, I ended up buying and it was indeed a good call. Then I needed a low profile cooler for it, and it recommended a Thermalright AXP90-X47. I am still about to receive it, I think it was a good call because it filtered hundreds of options out there and gave me what apparently fits my needs, budget and limitations. I assume another successful purchase suggested by AI that helped me narrow my choices.

Maybe I got lucky? Does any of you ended buying something clearly wrong because of bad advices from AI?

Hi,

Funny I started a chat with Gemini and then ChatGPT about AVP and provided as an output
minidsp Flex HTx
Topping DM7
….

My ask was I want to avoid buying a dedicated AVR instead does it exist a dac with multiple output for video (in short story).

I’m using AI to very much narrow my choice with interesting explanations.

 

[Brian Hall]

I used two locally installed and free to use AI Models, Gemma4 27B A3B released by Google and Qwen3.6 27B released by Alibaba to research attic cooling. Both of these local AI models run on my Linux pc with LM Studio (also free). There is a version of LM Studio for Windows if you haven't upgraded to Linux yet. The AI models are just data files so the same models can be used for either.

Both AI models were correct in recommending Radiant barrier first, and a then a gable exhaust fan. They also said to make sure the existing soffits were adequate (they weren't so I replaced some and added 6 more at the north side of the house).

The combination lowered the attic temperature from 150F to ~90F on days where the outside temperature was over 100F. It is currently 104F outside and my attic temperature is 92F. Removing all of that extra heat from the attic made it much easier and less expensive for my AC to cool the house.

 

[Berwhale]

I recently got the 'upgrade itch' with my PC. There's nothing wrong with it specs-wise (Ryzen 9 7900, 32GB DDR5 6000 RAM, 2x 2TB PCI-e 4 NVMe, Radeon RX 7800XT), it's just 4 years old now. So I decided to 'upgrade' my perfectly good CPU cooler (Thermalright Peerless Assassin 120) with an AIO watercooler.

I'd been considering the Arctic Liquid Freezer III 240 for awhile when a good deal came up on Amazon UK (£56.49), so I bought it, installed it and gave the inside of my PC a jolly good clean with an air duster and vacuum cleaner.

The Arctic Liquid Freezer III 240 is a very good watercooler, the problem was that my case, a Fractal Design Pop Mini Silent, was designed to be quiet. The case came pre-fitted acoustic foam and has minimal gaps for air intakes to minimize noise leakage.

It turns out that two 120mm fans pulling air through a 240mm radiator on the front of a Pop Mini makes a very annoying whistling sound, even at relatively low RPM. At this point, I could have put the Peerless Assassin back in and returned or sold on the watercooler, but then I would still have that itch!

So I asked Gemini Pro what the best microATX case would be for the Arctic Liquid Freezer III 240. I suggest several cases, including the non-silent 'air' version of the Fractal Design case I already owned. I ended going for an NZXT case on Gemini's recommendation (I got another good deal on Amazon UK - £49 )

Spoiler: Gemini Pro

When choosing a Micro-ATX (mATX) case for your setup, radiator thickness headroom is the single most critical factor.

The Arctic Liquid Freezer III 240 features an extra-thick 38mm radiator combined with 25mm fans (totaling 63mm thickness). Standard AIOs are only 52mm thick, so many standard mATX cases will run into motherboard VRM heatsink or RAM clearance issues with top mounting.

The top mATX cases that easily accommodate top-mounted 240mm radiators—specifically with enough clearance for thick coolers like the Arctic Liquid Freezer III—include:

Top Picks for Micro-ATX Cases​

1. Lian Li A3-mATX (Best Overall / Compact)​

• Volume / Design: ~26.3L mesh enclosure co-designed with DAN Cases.
• Radiator Support: Up to 360mm / 240mm top-mounted radiator.
• Why it’s great: Excellent spatial efficiency and layout versatility. Because the PSU mounts vertically at the front/side, the entire top panel has generous clearance, meaning the thick 63mm Arctic LF III fits with plenty of breathing room.

2. ASUS Prime AP201 (Best Airflow / Easy to Build)​

• Volume / Design: ~33L full-mesh or tempered glass chassis.
• Radiator Support: Supports up to 360mm top radiators.
• Why it’s great: One of the most popular mATX cases on the market. The top radiator rail is deeply offset from the motherboard tray, allowing thick 240mm and 360mm AIO combos to sit clear of motherboard heatsinks and high-profile RAM.

3. NZXT H3 Flow (Best Modern Airflow Design)​

• Volume / Design: 35L compact mATX chassis with fine mesh panels.
• Radiator Support: Up to 240mm top radiator.
• Why it’s great: Explicitly offers up to 70mm of clearance for top radiator + fan combos, ensuring the 63mm Arctic LF III fits without hitting motherboard components.

4. Fractal Design Pop Mini Air (Best Classic Aesthetic)​

• Volume / Design: ~36.5L traditional mini-tower layout with mesh front panel.
• Radiator Support: Up to 240mm top radiator.
• Why it’s great: Features a clean, dual-chambered aesthetic with pre-installed fans.

Important Note for Arctic LF III: While the Pop Mini Air supports top 240mm radiators, its top offset is tighter. Low-profile RAM (under 35mm height) and lower VRM heatsink heights are required to avoid physical contact with the thicker 63mm Arctic radiator assembly.

Case	Top Radiator Cap	Top Clearance	Ideal For
Lian Li A3-mATX	Up to 360mm	Very High (PSU re-positioned)	Smallest footprint, ultra-modular setups
ASUS Prime AP201	Up to 360mm	High (Offset rails)	Maximum GPU/AIO flexibility, clean mesh aesthetic
NZXT H3 Flow	Up to 240mm	Up to 70mm	Guaranteed Arctic LF III clearance, modern look
Fractal Pop Mini Air	Up to 240mm	Medium (Watch RAM height)	Traditional layout, stealth optical drive slot

 

[Matias]

It's not something I bought but recently I had AI review my workout. I used ChatGPT and told my age, weight, height, activities during the week and sent my complete program, each exercise, reps and weights. It analyzed everything and came back with suggestions where the training was already good enough and what was missing, suggested an exercise and even the images showing how to do it safely. Amazing!

 

[Zetti]

I get that AI can induce some bad things (laziness, incorrect information, slop, many other things). But it can be incredibly useful as well. So dismissing it outright makes little sense to me.

I have no AVR, and here in Brazil they can be quite expensive, especially one with multichannel pre-outs (I have active speakers). I wanted to get a little taste with multichannel (kinda, 4.0) audio for movies and TV shows, but could not spend money on an AVR. AI managed to make quite the gambiarra (not easy to translate this word, but I think context will make its meaning clear in a few seconds) with equipment I already had. Basically two DACs, each fed via USB from a computer running Linux, which handles all the routing and audio stuff. I used AI to setup this, and also correctly handle center channel, bass management, loudness adjustments, dynamic range compression for night-time watching. All via CamillaDSP. I know using two DACs is not ideal, but this is just a test. And it works. And I spend zero money on it, setup was quite easy and done in minutes. Without AI I would never even try.

Right, this is the complete opposite of what the topic was asking, but I wanted to share the experience.

(I also used ChatGPT to create a browser-based training program which I had on an electronic book on my computer. I track weights, reps, exercise selection. It also tracks the weekly progress. Very neat, for free, done in minutes, with very little back-and-forth needed.)

 

[AJM1981]

Found a great pair of reasonably priced multifunctional earbuds. Also managed to find a Tribit bluetooth speaker for someone, which is basically like what Bose produces, minus te costs for its snob status and marketing department. Well reviewed stuff.

 

[Chagall]

I bought my car insurance using AI.

Got six offers in total - three mandatory insurance and three casco/comprehensive coverage, and each one was something like a 10-15 page PDF.
Fed them all to GPT and asked it to compare them. Got a short pros/cons list with the meaningful differences highlighted. After a bit of back and forth, I had enough information to make a decision.

I'm totally fine with AI doing the grunt work instead of me.

 

[Axo1989]

Ok, ok, change that to "a lot of people". Also because now Google Search already gives an AI answer by default whether users ask it or not, so one has to actively ignore that part (or use another search engine).

If I hadn’t already stopped using Google’s web search for other reasons, I would certainly have stopped using it for that reason.

And no, I haven’t prompted a chatbot to output purchasing advice. Why support such things?

 

[Thighsofsteel]

I bought some excellent large PTFE furniture sliders. They have been very useful.

 

[Soundproof]

Nothing. Haven't purchased anything by AI advice - knowingly.

As my purchasing agenda mainly relates to audio these days, I take my advise from ASR/Eirin and so on. Then I look for the low ratings and negative comments to see if there are any potential caveats with the product, or the seller/distributor. It tends to pile up if there's actually something bad going on, or it can be ignored or irrelevant.

Even if not neccessarily purchasing the ASR/Eirin/++ products found to be references for what I'm after, they do serve as references - when making compromises, I like to be aware and deliberate with the compromises I make.

I somehow don't picture AI as a replacement for that process (and the fun of it), I prefer the sources to the mashup.

 

[maxmax]

I'm currently looking for different audio DSP solutions. AI helped me to find several products. I then continued to ask for recommendations and during the dialog I went back and asked critical questions about information AI gave me earlier. The funny thing is that the AI went back several times and stated things like "I have to correct my earlier statement, solution x is not superior to y...". So after all, AI helped me to stay away from a purchase, a real money saver

 

[Matias]

I am again building PC parts with the help of AI, not only if they work together, but better in a sense of if the part is "proportional" to the rest of the system, it is checking benchmarks and prices in milliseconds to tell me if it is worth it or not. And even when it is not, unless I find it used for X price, then it is worth it, which I followed. Impressive.

 

[Mnyb]

We booked train with the help of AI this summer we where going by train around Europe at our vacation , it was really good at compiling info about train schedules from different sources

 

[Cote Dazur]

Never did but not saying never will, AI will be part of my future, that I want it or not.