Robert Carnes

I worked with AI to research and design a guide to some new (and old) terms around AI phenomenon. There are plenty of good ones on here, but "scuffing" might be my favorite new one I learned.

Go into the memory of the LLM tool you use and check out the notes it takes for itself about you. The things it "remembers". You can also edit this memory. Meaning you get to incept the tool to make it think things about you.

Alan Tudyk is the secret engine that keeps modern animation going. He's such an underrated voice actor.

One of the cool things about AI is the compiled history of random conversations, organized by theme. A huge problem is how poor the search feature for LLMs is. Why aren't their search functions better infused with AI? Huge missed opportunity.

Why isn't there a medical first responder in the paw patrol? You have these kind of randomly specific observations when you watch kids shows as a parent.

To everyone who is worried about "missing the boat" on AI, not every boat is worth catching (see: Titanic). I'd rather be behind the curve that running blindly towards a cliff with the other lemmings.

It's only slightly dissapointing to me that the word "oroborus" isn't a pallandrome. (Neither is pallandrome for that matter.) Seems like a missed opportunity.

16 Sept

I've started counting how many flock cameras I see when driving around town. There are 6 on my (short) commute to work. And I counted 11 driving a few miles to Costco.

15 Sept

Claude keeps trying to suggest next steps on queries, and name dropping colleagues I should follow up with. Except it's usually wrong, and always annoying. I keep asking it to stop. Then it apologizes and says it'll stop. And then keeps right on doing it. AI is not getting smarter.

11 Sept

Watching Roy Kent jump around with Jamie Tartt was the highlight of the entire new season of #TedLasso so far.

4 Sept

I love the frankness of this @wikipedia.org disclaimer. Maybe some of the article's multiple issues stem from the fact that it's an orphan. Wikipedia articles have feelings too. (Right?)

1 Sept

It looks like Asana is now requiring for you to have an enterprise level account to opt out of AI features. If these AI features are so great, why are they being forced on us?

1 Sept

There's a big disconnect between the companies who keep trying to cram AI into every component of their products, and we the consumers who are weary and distrustful of anything AI. I'd love for just one tech company to decide not to use AI.

21 Aug

I originally subscribed to Disney+ when it launched in 2019. It cost me $6.99/month then. It's now up to $32.99/month. Now, that includes the Hulu/ESPN bundle. But the bundle was originally $19.99/mo. The + must represent the addition of extra pricing.

19 Aug

I was reading this article about the increasing threat of AI in cybersecurity. Until it became painfully obvious it was written by AI. Interesting that the by line is "edited" by a human. So they're not even denying that. Just a tad ironic.

18 Aug

Started messing around with Canva's AI feature. So far it's laughably useless. In case you're curious.

17 Aug

I love that @johngreensbluesky.bsky.social, Brené Brown, and John Oliver are @lfcr.bsky.social fans. Three people I already liked who I now like even more.

14 Aug

I can't recall ever buying a home printer and not being disappointed by it.

14 Aug

Not everything needs a mobile app.

13 Aug

I wish people would say that AI is "improving" or "getting better." From my subjective perspective, they're getting worse. Because AI tools have net cost me more productive time than saved time. Several times, I've had to fix stuff that it's messed up.

12 Aug

If Dune 3 & Avengers: Dooms Day are coming out on the same day in December, we're calling it Dunes Day, right?

31 July

Everyone stop suggesting adding AI-powered chat to websites. They're fine at regurgitating the info in your FAQs, but that's about it. Any actual questions or deeper troubleshooting, they're useless. But there's no person to contact, so it's a net negative.

31 July

Since posting this, I've interacted with two more AI chat bots. They both got caught in loops suggesting the same advice that didn't solve my problem. Every one I've interacted with has been roughly the same. Which is to say, maddeningly unhelpful.

31 July

Google Search Console now allows you to connect social platforms for search visibility on each. This is cool for digital marketers and also aggrivating for people who care about data privacy.

30 July

Question: how sure are we that the advertising metrics that Meta and Google provide are accurate? Because there are times when I review the numbers that they don't seem to be accurate. And there's no real way to verify.

30 July

Imagine a world where they've built trillion dollar businesses on the backs of digital ads that are compeltely made up and arbitrarily inflated. Not saying that's the case. But what proof do we have that it's not? They could easily inflate this data and we couldn't tell.

30 July

There's clearly some inherint value in these ad platforms, or businesses wouldn't invest in them. But when we lose trust in the veracity of the data, their percieved value vanishes very quickly.

21 July

After 2-3 months of exclusively using Claude, I tried to use ChatGPT for a simple image generation task this morning. It couldn't do it. Kept running into errors. And very slow. It's gotten so much worse than it already was. Why does anyone use this tool?

8 July

A coworker got a new PC laptop. It has an "AI" button that automatically opens Copilot. It's where the old CTRL button was, which she used a lot. She had to go into settings to disable the useless button to something useful. People don't want AI as much as tech orgs think we do.

Something we'll have to grapple with as we enter the future of work with AI tools: What to do with the lag time as an LLM processes an output. Or: what do you do when AI is working on something?

Keep in mind, this is also a co-worker who has very strict (but vague) limits on how much work they can do in a set period (token caps). So you also have to wait (sometimes hours) for a limit to reset, or pay through the nose for usage overages.

Just reflecting on the massive "second-order problems" that arise from AI in work. Not enough people are talking about this stuff. It makes work far more complicated than we're being led to believe.

Something we'll have to grapple with as we enter the future of work with AI tools: What to do with the lag time as an LLM processes an output. Or: what do you do when AI is working on something?

The challenge is not knowing how long that time will take. And not knowing how deep to get into your own work before it's finished and pulls you back out. Basically, this is another thing people can "multi-task" at, which is really just productivity-killing task switching.

We can imagine treating this like delegating an assignment to a colleague. What did you do while you waited for a co-worker to get back to you? Also, would you let them interrupt you when they finished, and force you to drop everything you're doing?

Something we'll have to grapple with as we enter the future of work with AI tools: What to do with the lag time as an LLM processes an output. Or: what do you do when AI is working on something?

People want to pretend like AI is "so fast" at doing everything. But the reality is that it still takes time. Plenty of the complex tasks I give Claude take 10-15 minutes to run and spit out. So what do you do in the meantime with that time?

You know what's even better than watching the World Cup? Watching groups of fans react when their country scores a goal. It's riveting.

Stop saying: "This is the worst AI will ever be. It's only going to get better." I understand the argument, but fundamentally disagree. Especially because "better" is so wildly objective.

Primarily because of the corporate aspect. AI might become more capable, but so will companies of exploiting it for profit. We could say the same "only get better" statement about the internet as a whole. Which has objectively gotten worse over the last decade.

So much of our current AI hype is built off faulty logic that AI companies encourage us to buy into so they can avoid accountability. It very well could get better. But not if we just assume it will and then abdicate our responsibility to make sure it does.

29 June

Dear website companies, Despite what you think, I do not want to interact with your website's AI chatbot. If your site requires an LLM to navigate, then it's just a bad website.

29 June

Seeing a bot on your site makes me trust you less. It's like an annoying pop up that provides me no added value. If I want to use an LLM, I'll use my own that should be able to pull info from your site. Not some generic one you've provided.

It's also clear that so many support chatbots are just general LLMs that have very limited knowledge of their own company's product. I'm chatting with one now that's just guessing. It keeps asking me "tell me what you see" rather than knowing how the product works.

Why are home printers so awful? We got a new $100+ HP printer recently. I had to download an app, create an account, sign up for a membership to get ink... ...and yet I still can't get it to print anything.

Sure AI doesn't sleep or need to eat, but it does crash spectacularly at the most inconvenient times. How can we build our entire economy on something that could suddenly stop working for hours at any point?

Related, I wanted to figure out how much time each year 99.12% uptime is. But I'd normally use Claude for that, which I can't now. Just saying: we should NOT put all our proverbial eggs into this unstable basket.

OK, I did the math on my own and that's almost 4 days of downtime every year. And this is on one of our more reliable models. That's massive.

essays.highagency.comThe Adults Don't ExistCure yourself of adult syndrome

16 June

Adult Syndrome is something I've felt deeply for a while now. Kudos to George Mack to putting words behind an important, but difficult to articulate concept. essays.highagency.com/p/adults-don...

15 June

Google has now fully injected Android Auto with AI. Now, all the things I used to do via voice: send a text start a call with a contact play a podcast get directions ...are now all completely useless. I've tried so many times, and the response is always: "Sorry, I can't help you with that."

12 June

A reminder text from my dentist started with: "Here's a more polished version of your message..." So clearly they were using AI to text me and didn't bother to check it. And the message was just: "We need to reschedule your appointment..."