Just because a part has the shape of an engineered part does not make it compatible, strong, safe, and fit for purpose. This part could have likely been fine if it used a different material such as Ultem.
According to the report:
> The aircraft owner who installed the modified fuel system stated that the 3D-printed induction elbow was purchased in the USA at an airshow, and he understood from the vendor that it was printed from CF-ABS (carbon fibre – acrylonitrile butadiene styrene) filament material, with a glass transition temperature3 of 105°C.
https://assets.publishing.service.gov.uk/media/69297a4e345e3...
Isn't this simply a part that shouldn't have been allowed to be sold based on it being both faulty and also misleading?
And if this part was simply 3d scanned and printed in whatever material seemed strongest,
Then it could be an apt analogy
Yeah, exactly -- which is why it's a stupid phrase for what happened here.
Not every negligence is somehow equatable to an AI pitfall, it's just on parents' mind so it's the only metaphor that gets applied.
A poorly fit hammer in a world of nails.
I say this as an engineer/proprietor with years of additive manufacturing experience, it's insulting. A poorly chosen and wrongly used process conveys nothing about the underlying fundamentals of the process itself -- it conveys everything about the engineer and the business processes that birthed the problem.
Similarly if I came across a poorly vibe-coded project I wouldn't blame Anthropic/oAI directly -- I would blame the programmer who decided to release such garbage made with such powerful tools..
tl;dr : it's not vibe-coding itself that makes vibe-coding a poor fit to rocket science and brain surgery -- it's the braindead engineer that pushes the code to the THERAC-25 without reading.
The comparison does not seem as absurd to me as it does to you. vOv
I see multiple examples of it in this thread.
Analogies can be useful sometimes, but people also shouldn't feel like they need to see everything through the lens of their primary domain, because it usually results in losing nuances.
(unless that primary domain tends to attract a lot of people who tend to the hyper-literal /s)
And since vibe coding is so recently coined, I think a lot of people take it to specifically mean "LLM" and not some generalized "any third-party agent".
Then, a vibe coded engine part sounds like it would need a generative AI producing the CAD file that is then printed. And it might have some bizarre topology like a Klein bottle or some fever dream.
1: https://nymag.com/intelligencer/2014/10/soylent-creator-hack...
"Faulty and also misleading" pretty much describes the code output by vibe coding. And don't get me wrong: I use LLMs daily to help me write functions / write test cases / find bugs / find edge cases / explain error messages etc. But I don't vibe code entire parts like that part that melted.
With the rise of accessible 3D printers that can print engineering materials, there are a lot of people who try to create functional parts without any engineering background. Loading conditions, material properties, failure modes, and fatigue cycling are all important but invisible engineering steps that must be taken for a part to function safely.
As a consumer with a 3D printer, none of this is apparent when you look at a static, non-moving part. Even when you do start to learn more technical details like glass transition temperature, non-isotropic strength, and material creep, it's still not enough to cover everything you need to consider.
Much of this is also taught experimentally, not analytically - everyone will tell you "increasing walls increases strength more than increasing infill", but very few can actually point to the area moment of inertia equation that explains why.
3D printing has been an incredible boon for increasing accessibility for making parts in small businesses, but it has also allowed for big mistakes to be made by small players. My interpretation is the airshow vendor is probably one of these "small businesses".
Looks good - falls apart in practice, and a junior can't tell the difference as they "look the same" to the inexperienced eye.
From practical experience, you cannot just replace a tyre on a car with any old bit of wood - you really need to use hard wearing mulga (or equivilant) as an emergency skid. (And replace that as soon as possible)
This whole thread is a stretch, IMO. But, I like this phrase.
As a fabricator (large wood CNC, laser cutting and engraving, 3D Printing, UV Printing, Welding). I put engineering into a whole different job scope. I can make whatever you tell me really well, not vibe-carving.
I don't necessarily write the specs or "engineer" anything. I'm just saying, don't blame the medium, 3D printing. The fact is a fabricator is not necessarily an engineer, regardless of the medium.
Using scrublands wood (slow growing tough long grain mulga) as a skid when a rubber tyre destroys itself is an old old hack passed on by my father (he's still kicking about despite being born in the early 1930s).
In the early 1980s I used to enjoy hanging out with Chris Brady and helped out making jigs to assemble snare drums: https://www.youtube.com/watch?v=jdBHtUN5gAE
His jarrah, wandoo, and sheoak snares are still loved: https://www.youtube.com/watch?v=tKmDuu5Iba4
Point being, I don't blame processes (3D printing, etc) for part failure, that comes down to whether the shape and material are fit for purpose, whether material grain structure can be aligned for sufficient strength if required, whether expansion coefficients match to avoid stress under thermal changes, etc.
Engineering manufacturing can sometimes be suprisingly holistic in the sense that every small things matter including the order in which steps are performed (hysteresis) .. there's more t things than meet the eye.
Everything you need to consider is really not that much when it comes to most typical consumer 3d printing projects. Mostly because they are usually about stuff like "fixing a broken tashcan". The engineers who made that bullshit plastic part that broke after a year probably knew all about area moment of inertia, but that doesn't mean I need to to print a replacement part that lasts longer - or not, in which case I'll just iterate on my process.
I really don't get the dismissiveness, and frankly, I've never experienced that from engineers in my life. They just seem delighted when someone, kid or adult, tinkes with additive manufacturing.
I think both vibe coding and 3D printing are wonderful things. Lowering the barrier to entry and increasing technology accessibility allows those without formal training to create incredibly capable things that were previously difficult or not possible to do.
What I meant to specifically highlight is the 3D printing of functional parts that have some level of impact on safety, things that can lead to significant property damage, harm, or loss of life. Common examples include 3D printed car parts (so many) and load bearing components in all sorts of applications (bike mounts, TV mounts, brackets, I even saw a ceiling mounted pull-up bar once).
This isn't to say it can't or shouldn't be done. What I'm saying is that both on the digital side (files for personal use) and the production/sale side (selling finished parts), there is no guarantee of engineering due diligence. 3D printers enable low volume small businesses to exist, but it also means that, purposefully or not, their size means they can go quite a while without running into safety regulations and standards meant to keep people safe.
Not a new story in the progression of human endeavors; see the printing press, perspective painting, digital photography, residential construction.
The vendor selling the 3D-printed part at an airshow probably didn't think: "I'll deceive pilots." They likely thought: "I can 3D-print this part to spec, it looks right, it fits, and pilots will be happy." The capability to create professional-looking outputs outpaced the discipline required to validate them.
Same with vibe coding: the LLM isn't lying. It's producing code that passes basic inspection. But both technologies have collapsed the cost of creating something that looks production-ready while preserving all the ways something can fail in actual use.
Before 3D printing and LLMs, there were natural friction points that forced validation:
• Manufacturing a metal aircraft part required industrial equipment, precision tooling, material selection expertise. The process itself embedded quality gates.
• Writing professional software required years of training, code review practices, deployment infrastructure. The difficulty forced rigor.
Now, both technologies let anyone produce outputs that visually and functionally resemble professionally engineered work without any of the underlying validation:
• A 3D printer can output a part that looks dimensionally correct, has proper tolerances, fits perfectly—but the material choice was never stress-tested against thermal cycling.
• An LLM can generate code that compiles, runs, produces correct output for test cases—but has no error handling, SQL injection vulnerabilities, or memory leaks that only appear at scale.
This is a relatively new failure mode where professional appearance becomes decoupled from professional rigor. And customers can't easily tell the difference until something breaks.
Also, the part it was replacing was a fiberglass part in an epoxy resin, with a glass transition temperature of 84°C. So the 105°C glass transition temperature of the replacement part should have been better than the original.
However, the original had an aluminum tube supporting the inlet, which provided extra structural support beyond fiberglass epoxy resin. And upon testing, the actual glass transition temperature of the 3D printed part was 52.8°C for one sample and 54.0°C for another, so much lower than expected.
Now, because the regulations are much less strict for experimental homebuilt aircraft, there may not be the traceability to figure out where in the chain the issue came up. Was it a bad batch of filament? Did the person making the part use the wrong kind of filament? Who should have tested the glass transition temperature of a coupon of the same material as the replacement part? Did the 3d printed material glass transition temperature change over time, possibly due to something like fuel or exhaust fumes?
The recommendation from this report is to disallow 3d printed replacements for this part, but it should be possible to do with the right material and proper testing and analysis (as well as leaving in the aluminum tube for additional support), as this is an air intake and it should be possible to find a 3d printed material that can withstand the kind of temperatures an air intake is subjected to, given that the original part is a fiberglass with epoxy resin.
Installing life-critical parts of shoddy engineering into a vital system of your airplane is a good example of when things do matter.
That should be so obvious that I wonder if it was DIY by the pilot.
> The Cozy Mark IV is a 4-seat, single engine, homebuilt light aircraft [...] The aircraft is built from plans using basic raw materials. It is not a kit aircraft
You could scarcely get more DIY than this aircraft. Home-built, and not even from a kit - the builder gets to lay up every part in glass fibre themselves, by hand. And this guy had been flying it for 26 years.
It sounds like the guy was sold a part 3D printed in the wrong plastic, and it melted. He thought it was ABS, but it melted at the temperatures PLA melts at. If your engine air inlet is made of plastic that melts at 54°C (130°F) you're going to have a bad time.
It's easy to imagine how a chaotic 3D printing business might have run off a test part in a cheaper black plastic, then a confused worker could have stored the test part in with the other 'identical' parts in a different black plastic.
The 'serious' aerospace industry avoids this with lots of paperwork and procedure; when an airline maintains an airbus plane, they use only airbus-approved parts from airbus-approved sources with a paperwork trail confirming they were inspected for being-the-right-material using an approved procedure. I don't know if the home-built aircraft community would be eager to adopt those practices, though.
I don't know how the regulatory environment is in the UK for experimental craft (this is considered to be "experimental" category in the US and Canada), but yes, the idea behind an experimental is that everything is DIY.
I have an experimental, and I can do close to anything I want. What I can't do is complain when my plane crashes because I installed a part that isn't fit for duty. I, as the owner and operator, am the one that signs off on the airworthiness of the plane.
E.G. If I install a Cessna part on my plane, and that is the cause of a crash, that is my fault from the point of view of the FAA.
There may be legal considerations outside of airworthiness and flight rules, but as far as the FAA is concerned (or would be if this had happened in the US), the manufacturer of a part is off the hook once the thing is installed on an experimental.
Maybe, but FDM printed parts are still much weaker than molded parts. We tried printing some coolant pump housings once during development. They worked fine until the pressure went up and then layers separated and someone got to clean the lab. At least an air intake is gonna have negative pressure which might help hold the layers together.
Looks like they would like to make the early flight mistakes themselves instead of following air worthiness guidelines.
In the end it depends on the application. Vibe coded flight management systems, anyone?