Tag: ai
-
Today, an AI review tool used by my workplace reviewed some code that I wrote, and incorrectly claimed that it would introduce a bug because a global variable I created could "be available to multiple browser tabs" (that's not how browser JavaScript works).
Just in case I was mistaken, I explained to the AI why I thought it was wrong, and asked it to explain itself.
To do so, the LLM wrote a PR to propose adding some code to use our application's save mechanism to pass the data back, via the server, and to any other browser tab, thereby creating the problem that it claimed existed.
This isn't even the most-efficient way to create this problem. localStorage would have been better.
So in other words, today I watched an AI:
(a) claim to have discovered a problem (that doesn't exist),
(b) when challenged, attempt to create the problem (that wasn't needed), and
(c) do so in a way that was suboptimal.Humans aren't perfect. A human could easily make one of these mistakes. Under some circumstances, a human might even have made two of these mistakes. But to make all three? That took an AI.
What's the old saying? "To err is human, but to really foul things up you need a computer."
-
claude, make it make sense *harder*
blackle mori delivers a welll-rounded, one might say voluptuous, take on the writing process, glistening with the fiery passion of its author. I desperately want to get off... how fun this post was.
-
Hello World by Hannah Fry
It took me a while, but I finally got around to reading Hannah Fry's "Hello World: How to Be Human in the Age of the Machine", and it was really good!
-
AI and cigarettes
Chris Ferdinandi draws parallels between people pushing 'less-harmful' AI and those promoting vaping as 'less-harmful' smoking; the latter, of course, ultimately resulted in new people taking up a nicotine habit that might now have otherwise: not what was intended at all!
-
We Need to Talk About Botsplaining
In a wonderful blog post from last month, Stephanie describes the scourge of Botsplaining. Here's my experience of it.
-
AI assistants misrepresent news content 45% of the time
The BBC and EBU did some research and discovered that AI assistants tasked with bringing you news get it right only slightly more-often than they get it wrong, and they get it catastrophically wrong about one time in five. This should be a surprise to nobody.
-
Somebody just called me and quickly decided it was a wrong number. The signal was bad and I wasn't sure I'd heard them right, so I followed up by replying by text.
It turns out they asked Siri to call Three (the mobile network). Siri then presumably searched online, found Three Rings, managed to connect that to my mobile number, and called me.
If Siri's decided that I represent Three, this could work out even worse than that time Google shared my phone number.
-
The real (economic) AI apocalypse is nigh
Cory Doctorow paints a bleak picture of the future world economy in the aftermath of the bursting of the AI bubble. I agree that we're in a bubble, but I'm slightly more-optimistic that we can survive it without a total economic catastrophe.
-
Will Smith’s concert crowds are real, but AI is blurring the lines
A recent video by Will Smith gained widespread condemnation for using AI to artificially-inflate his crowd's size. Except that's not what happened at all: a little creative editing by his media team, combined with a secret experiment by YouTube to make some videos look immeasurably more-shit by performing AI upscaling on them, seems to have kicked off this particular round of Internet craziness.
-
Four perspectives on AI
Sean McPherson shared a "four quadrants" diagram for people's beliefs about AI, with a specific focus on LLMs used for coding assistance. Here's where I sit on it, and why.
-
Just a Little More Context Bro, I Promise, and It’ll Fix Everything
Based on my experience of 'vibe coding', I think Jim Nielsen's analogies are absolutely spot on.
-
The rise of Whatever
A freaking excellent longread by Eevee (Evelyn Woods), lamenting the direction of popular technological progress and general enshittification of creator culture. It's ultimately uplifting, I feel, but it's full of bitterness until it gets there.
-
ArtificialCast
ArtificialCast is a lightweight, type-safe casting and transformation utility powered by large language models. It allows seamless conversion between strongly typed objects using only type metadata, JSON schema inference, and prompt-driven reasoning.
-
Adding a feature because ChatGPT incorrectly thinks it exists
ChatGPT lied about the functionality of Adrian Holovaty's software tool, Soundslice. So rather than try to get ChatGPT corrected, he just... implemented the feature it had claimed he already had. Is this the first time in history that this has happened? It could be!
-
The Huge Grey Area in the Anthropic Ruling
The court ruling that declares Anthropic's use of (purchased) books as training material leaves a curious sorites paradox. If we assume that an LLM trained on only one book (and so only capable of producing content directly inspired by that book) would necessarily be copyright-infringing, at what number of books does that cease to be the case?