It's puzzling, considering they made some strides towards improving their design language and systems in past years, which produced IBM Plex etc. under Mike Abbink.
They have a decent component system with Carbon[0] and a whole design principles website[1] which they could have easily pointed their LLM of choice to.
Even the ibm.com landing page isn't close to the design system spec, rounded corners and whatnot...
> M.R. Williams is advancing its IBM i modernization strategy with new mobile solutions and accounting system improvements, while leveraging IBM Bob to quickly interpret older RPG code, automate documentation, and streamline development workflows.
This is exactly the kind of endorsement I’d expect to see from an IBM customer.
Yes. I missed that specific bit, but, the more I read on, the less it felt like a nice in-flight magazine-tier diversion and the more it felt like I was reading the result of asking Claude to write an essay about the bearing of loads.
(Which makes me personally a lot less interested in reading it.)
Here's a certified human written post from the one and only Yvon Chouinard about tumplines. Patagonia actually used to sell a tumpline compatible with some of their packs briefly in the 2010s!
Yeah the writing tastes like processed food. There’s no substance just “here’s some stuff people do” but nothing about why it works or where you could learn more.
I suppose I read enough books that I've seen this from human writers all the time for decades. It's even an English construction that I was specifically taught Japanese and Spanish equivalents in my foreign language classes decades ago, and I use the German equivalent with my family all the time.
Assuming it is AI-written, who cares? Would the article have been written if AI hadn't authored it? Did you derive less utility from the article than had it not been written at all?
Bunch of Veruca Salts here. I want ALL the entertainment/information, I want it NOW, I'll pirate it if I have to, and GOD FORBID IT USE AI.
The intent to communicate particular ideas is actually important to the writing. If you have a lot of well put together words with relatively low intentionality behind it, what you have is something that is very effective at wasting your time. If I want to waste my time I'd rather be playing videogames.
Indeed! And even subjectively, I find it difficult to believe that even if AI-written, the new information learned has negative marginal utility (or less marginal utility than had it been human-written). That is in fact the basis of the point I am making, though it is implied rather than stated.
I care. If the article was AI written, I'm massively less interested in writing it, and in fact I'll only do it if absolutely necessary, or if I'm getting paid.
That's just how it is for me. If you disagree: no problem! Some people are inclined to argue with people about their opinions, but I'm not one of them, at least not on this matter.
The messenger tells you a lot about the message with AI articles.
It being written by AI instead of a human is an indication the writer didn't care enough about the information being communicated to communicate it themselves and if they didn't care enough to put effort into writing it and making sure it is actually good why should I spend my time reading it?
It's also an early indication that the content won't be that good too so it's a convenient signal that reading it will likely be a waste of time. Just look at the article here. Lots of fluff and some completely vapid praise of some of the carry systems that put a significant load into the necks of the carrier. We know that's not good for you it's just convenient when you have low tech.
While I agree that the topic is actually pretty interesting and any decent info > no info, in more general terms we pay with the most important commodity we have - our free time.
So just because something is free doesn't mean people automatically flock in masses for it and have uncritical relationship with it.
I care. I'm interested in what humans have to say, not the slop produced by a clanker.
> Did you derive less utility from the article than had it not been written at all?
Unironically yes. When I read an article and realize it's slop partway through, not only have I wasted my time but I experience the annoyance of having wasted my time. It would be better for me if such articles did not exist at all.
> Google DeepMind tested this impact by serving a model that used watermarking to a portion of their Gemini traffic and comparing thumbs-up and thumbs-down ratings. They found no statistically significant differences from the unwatermarked model. And in a controlled study, human raters comparing watermarked and unwatermarked answers side-by-side saw no difference in quality.
For some reason I had assumed testing this would be more sophisticated than just checking the thumbs up/down stats and user "vibes"
It's not just checking user thumbs up/down. As your quote says, they also did a controlled study with people rating the results. What else would you want them to do? The whole point is that it needs to introduce a detectable statistical difference, but humans should not be able to perceive it as a quality difference.
Retest on benchmarks whether it accomplishes tasks with the same success rates. Prose is only one thing.
Messing with the randomness may make the problem solving capabilities weaker. Probably it doesn't but this is the answer to what else I would want them to do.
> human raters comparing watermarked and unwatermarked answers side-by-side saw no difference in quality
Yes - maybe saying "vibes" was minimizing the effort but what I am trying to say is that even the controlled testing is just asking users whether quality is impacted or not. Which is subjective and thats what I meant by when I said "vibes"
Don't get me wrong - I have no idea how one would go about testing this with other methods; I was just stating my assumption.
Since they rolled this out to all users I had assumed there would be other testing involved.
I'm not a mathematician but to me it doesn't seem so far-fetched to think that there might exist some mathematical proof that ensures the indistinguishability
I believe typing is probably one of the most important skills in 21st century for knowledge workers and yet one of the most neglected. Hoping to change that.
Anyone can learn to type fast, without having to think about or look down at the keyboard. We are building the best tool to help people get there quickly
I feel the challenge of typing applications is boredom. It's hard to keep using them. The only ones I kind of keep going back to are games. There serveral modern adventure games and a real time strategy game (I don't remember their names) and then there's Typing of the Dead.
Typing applications has been an extremely crowded space since all the way back in the Mavis Beacon days - good luck!
Small bit of feedback that might help it stand out more from the crowd and appeal to HN users: Add in training for Colemak, Dvorak, and maybe regional variants like AZERTY.
I definitely approached it from an engineering perspective - building what I thought was missing in the space that I wanted. In hindsight, it may not be the best business to quit a FAANG job for haha.
> Colemak, Dvorak...
We have these :). Have QWERTZ and British QWERTY as well. Going to add AZERTY next and a few others.
I an also exploring more uniqie/niche ones like Kinesis keyboards, etc.but no plans for this just yet.
Haha - well hopefully coming from a FAANG you've got some financial runway at least :)
Some more feedback
I'm a fairly fast typist (~120 wpm) so I didn't dig too deep into your tool, but I did see that you seem to have a metric around highlighting for letters where people might be making mistakes.
You might already be doing this, but my back-of-the-envelope thought is that you really don’t want to think about typing in terms of individual letters; you want to think in terms of clusters. Ideally, you’d use some kind of frequency corpus or Markov model or something similar to break text into commonly occurring constituent clusters, e.g. the -ing in a gerund.
It's essentially the equivalent of "memory chunking" but applied to typing analysis. That’s not to say letter-level analysis is not useful (especially for hunt and peckers learning where the keys are) but as the person progresses, you’d want to introduce this idea of statistical feedback on common sequential clusters.
It's a personalized & data-driven typing application that targets your weak points
reply