I can clearly tell just by using websites and mobile apps, the amount of silly bugs in the last 3 years or so is ridiculous. QA roles are going down and everyone believes they can vibe code production grade products without meaningful knowledge.
"AI just weaponized existing incompetence." - Love this phrase, it summarises very well what AI is doing to software and it will get worse.
I know that some companies do this and I agree that the ones that do it, shouldn't. The trouble is that it's not possible to fairly assess these cases in a format like this.
I have been using Phind for building SQL queries and it's actually great to learn more about SQL, and the results are mostly correct. Just try to ask it to build some advanced query then you discover you don't truly know SQL.
I actually did just the same thing the other day, and it managed to make a fantastically complex query (by my standards) with relatively little fuss. It made quick work of building an entire php/slim api quite nicely as well. You need to know what you are doing and correct some mistakes related to specific implementations (in my case SQLite specifics and the Psr library I was using in Slim - not sql or php in general), but overall a pleasant experience.
This is a really project and really helpful to understand history.
I noticed that several data points about the Portuguese Colonial Empire are wrong, is there any place where I can submit a ticket about it?
When the royal succession crisis took place in 1580, according to the blood line, the King of Spain was indeed the next in line but both Kingdoms remain independent, you can also find evidence of this in the name: King Philip III was called King Philip I in Portugal, the following one (Philip IV) was named the Philip II. In Timemap, when you check 1580, it shows the Portuguese territories with the Spanish royal flag, which is wrong because everyone understood back then that if Spain tried to dictated anything about the Portuguese overseas territories, this would be taken as a declaration of war. This is reason why the Treaty of Tordesillas was signed, Portugal and Spain divide the world and would not step on each other.
Also found things like Malacca, the flag is missing the dates of duration: 1511-1641
Same for Macau, the map states that the Portuguese rule ended in 1845 but in reality it only became independent in 1999.
Many other important missing bits that, although technically they don't as territories, do represent groups, example: The city of Nagasaki was built/shaped by Portuguese merchants during 1511-1641 and was indeed under Portuguese administration during 1580-1586.
Among many other bits that would make this reply too long for HN.
Words can not express my hate for this kind of articles.
Imagine working on a legacy codebase where the PM holds the dogma of refactoring being a bad thing and expecting you to do it wrong, even micro managing your PRs.
Most often than not, I do see projects suffering and coders actually resigning due to a lack of internal discussing about best practices, having space/time to test potential solutions, having Lead devs who resemble dictators quite well.
Let me guess, some PM wrote this article and they just want you to push the product asap by applying pressure and not allowing you ever to refactor. This is just a casual day in software development. I'm not surprised anymore when most web apps have silly bugs for years because it's gonna be a Jira ticket and a big discussion about..... one evil thing called refactor.
Several years ago I rewrote a full SaaS in about 3 months, it took another team 12 months with 5 devs. Guess which version made the investors happy, mine.
Bad refactoring is just a product of poor engineering culture.
> Imagine working on a legacy codebase where the PM holds the dogma of refactoring being a bad thing and expecting you to do it wrong, even micro managing your PRs.
I don't think the article said that anywhere? It was just a list of some common things that can go wrong when refactoring, along with some examples.
Nah, judging from the ancillaries (domain name, links to other articles, ads, etc) of the article, it was some guy selling an "AI" code tool of some kind who wrote the article.
(Probably a tool with Magikal Refactoring Functionality built-in... For a price.)
Hallucinations reduce the success rate of AI workflows, which must be taken seriously.
Imagine a workflow with 8 steps where each step/agent has a 95% success rate, the success rate of this workflow is only (1-0.05)^8 = 0.66 ~= 66%. Not bad but not enough to replace humans yet (unless 66% makes you profitable).
The hallucinations/errors compound and can misguide decisions if you rely too much on AI.
Not enough to replace humans in most critical tasks, but enough to replace Google, that's for sure. My own success rate to find information on Google these days is around 50% by query at best.
Definitely in my mind! Actually with Copilot you already have pretty good suggestions on autocomplete or with a prompt. But I am not sure of what would be an appropriate UI for a full-AI backend generator.
"AI just weaponized existing incompetence." - Love this phrase, it summarises very well what AI is doing to software and it will get worse.