You are trained on the content of your life

by Leave a Comment

Reading Time: < 1 minute

AI can do some amazing things, and I’m using it more and more frequently. However, it has a major limitation — it hasn’t lived your life.

Jay Acunzo worded this well in a blog post about his “Idea Impact Matrix” where he said:

AI is trained on internet content.

You are trained on the content of your life, and nobody else has access to that.

What you have to offer is unique from literally any other human that has ever lived. Your viewpoint on things has been shaped your entire life, and the perspective you have is slightly different than anyone else.

This is a big reason I spend so much time curating my notes in places like Obsidian (and now Tana). AI can surface a lot of great things, but those tools know about every book I’ve read, articles I’ve enjoyed, meetings with others, blog posts I’ve written — the more I put into it, the better the outcome will be.

I suspect AI will be able to replace that at some point, but that’s not very close yet. Even when it does, it’ll never be able to capture everything that has gone into shaping who I am today, and the same goes for you.

Nobody else has access to what you do, so use it well.

Don’t use AI for that

by Leave a Comment

Reading Time: 2 minutes

There are a ton of great uses for AI, and more coming out every day. However, there are places that you can use AI that are likely not a smart move.

I’m not talking about using AI to cheat on your homework or for other nefarious purposes (though those are clearly bad too), but places where AI can cheat you out of your own efforts.

For me personally, I can think of three areas where AI could likely “help” me but I’m intentionally not using it.

Moving highlights

When I finish a book, the highlights are automatically available via Kindle. There are tools that can take your highlights and automatically move them into systems like Evernote or Obsidian. I choose not to.

For me, the act of copying each highlight and moving it over gives me a chance to reassess if I want to keep it, a chance to group similar highlights together, and a chance to link highlights in my notes to related topics. AI could do much of that work for me, but if I highlighted something with the intent to learn from it, doesn’t it make sense to touch it more often?

Adding content to Anki

Similar is adding new cards into Anki. There are tools to automate that, but why? I use Anki specifically to try to learn and remember new ideas, and the effort to create those cards is a great first step toward learning.

Lots of stuff on this blog

Generally speaking, AI can be a great help for bloggers. You can come up with ideas, write outlines, and even publish fully AI-written posts. In my case, though, those are all bad ideas.

I want to read more to find more ideas, which I often find by linking notes together. For the post itself, my time and effort to write and edit is precisely why I’m doing this at all. Using AI for any of that could be counter to my goals.

AI is awesome, and it’s becoming a bigger part of my life every day, but intentionally excluding it in certain areas seems like a smart thing to do.

Using Google as a leapfrog

by Leave a Comment

Reading Time: < 1 minute

We all know that the nature of using Google is changing. Between people switching to using AI instead of traditional search, and Google keeping more users on their site, it’s all changing very quickly.

While listening to the annual “State of the Apps” episode of the Cortex podcast, the hosts shared some personal use cases that really stood out to me. Both of the hosts are using AI for the majority of their searches, and only defaulting to Google in certain cases. From the show:

Myke Hurley: “I’m just using Google as a leapfrog as a way to get to the thing that I already know exists.”

CGP Grey: “I’m using Google to get me to a thing that I already know I want to go to; that’s the only thing I’m using search engines for now.”

They debated the differences between various AI tools (they consider Claude to be better, but ChatGPT is often more useful), but they use those tools far more than they use Google.

Granted, they’re both on the extreme side of the innovation curve and most people aren’t there yet, but it’s certainly trending in that direction. As more people turn to AI for answers, getting your business found on the internet will be increasingly tricky to do.

Be “the” choice

The solution is to do exactly what Myke and Grey hinted at — make yourself the “thing that I know I want to go to“. If people are searching Google for businesses like yours, you have a chance to maybe show up. If they’re searching for you specifically by name, you’ve already won.

Be the choice.

Random posts

by Leave a Comment

Reading Time: 3 minutes

I’ve been slowly migrating my notes from Obsidian to Tana, and part of that involves essentially visiting every blog post on here again (inside of my notes) to get things properly tagged. It’ll be wildly useful when I’m done, but it’s rather tedious right now.

However, I’m getting a lot of value out of seeing those old posts again. There are a handful of old posts that I reference fairly often, but most posts are published and forgotten. This process has brought them back to light, and it’s been great.

I wanted to find a system kind of like Readwise that would take all of my posts and show me a random one every time I asked. As far as I can tell that system doesn’t exist, but please correct me if I’m wrong.

Instead, I had ChatGPT write me a simple WordPress plugin that does the trick, and it’s great! You can check it out here if you’re curious; every time you refresh the page, it shows another random post from my archive.

If I’m logged into the site, I also get a button on the screen that allows me to exclude that particular post from coming up again. There are a lot of posts on the site that are worth keeping on here but aren’t as valuable for me to revisit (like “new” WordPress features from years ago, or the various “Sunday Summary” posts). I’ll see them once, and then it’ll hide them going forward. The button just looks like this:

ChatGPT to write code

The experience of having ChatGPT write this code was mind-blowing. Here are the things I asked along the way in order to get what I wanted. Not only did it do a great job, but you can see my plans evolve as it came together:

  • Write code for a wordpress plugin that shows the full content of a random post each time the page is refreshed
  • This is good, but I only want the random post to show on a specific page, not on the main pages.
  • Excellent. Now I’d like to have it display the original post date on that page as well
  • Now include a button on the page that I can press that will exclude this specific post from appearing again in future random refreshes
  • When i click the “exclude this post from future”, have it load another random post once it’s done adding it to the exclusion list. Also, add a second button next to that one that simply refreshes the page to show a different random post.
  • When I click the “show another random post” button, the screen just turns white
  • It’s still showing a white page after I click that button (this time it solved the issue)
  • (for me, out of curiosity): Where are these exclusions stored in the wordpress database?
  • Only show the two buttons if a user is logged into the site

It was crazy to spend 10 minutes chatting with ChatGPT and getting a relatively complex plugin created. The refresh stuff I probably could have figured out on my own, but the “include a button on the page that I can press that will exclude this specific post from appearing again in future random refreshes” is way beyond my coding skillset these days.

This taught me a bit more about the power of ChatGPT, but it also gave me a page to visit from time to time to refresh my memories on those old posts. Win win!

AI detectors simply don’t work

by Leave a Comment

Reading Time: < 1 minute

Even though they’re widely known not to work, AI detecting software is still on the rise. Even those that work a little bit are getting worse every day as AI continues to improve. I saw a quote on Reddit that summed it up nicely, where a user said:

If AI could ever be used to distinguish between AI text and human text, that same AI would be able to generate human text. It’s a paradox.

Related is this podcast episode from Christopher Penn, appropriately titled “AI Detectors are a Joke”. I’ve done similar tests, running historical documents through some of them, and the results are about the same — most historical documents, like the US Constitution, are shown by these systems as being largely AI-written.

There are still huge challenges for how to handle the rise of AI, particularly when it comes to education, but using any kind of AI detection tool is a fool’s errand, and as Penn says, could lead to “potentially damaging results”.

I don’t have the answer, but these tools certainly aren’t are it.

The hopeful side of AI

by Leave a Comment

Reading Time: 2 minutes

There is a lot coming because of AI, and much of is not good. There’s not a ton of awful things to expect, but a lot of just… yuck. Like I shared yesterday, I suspect we’ll see more spam email, more AI-powered chatbots, more automated social posts, and just a lot of “we don’t need humans for that anymore”.

However, there are things to love about what AI can bring that go beyond just saying “we can be more efficient”. For example, here are three simple ones that we’re already starting to see:

Search and augment my notes
AI is doing some neat things with note-taking, among them being the ability to search and synthesize my notes, as well as add a bit more context to them. Being able to talk to an AI about my notes while I’m in them is pretty neat, and it will continue to improve rapidly.

Search my team’s notes
I thought we’d have this by now, but hopefully soon. For our team, essentially everything is in Google Workspace. We use it for email, calendars, documents, spreadsheets, video calls, Analytics, and more. I expect very soon to be able to ask “how many design comps did we promise Acme?” or “when did Sara and I discuss her challenges with MailChimp?” and get an instant answer. It’ll be awesome.

Medical care
I’m perhaps most excited about this, because the current and near future implications are giving us the best of both worlds; AI help, but with a doctor backing it so you don’t need to be scared of the outcomes. For example, AI is becoming quite good at identifying abnormal x-rays and when it finds a problem a doctor can pull it up to confirm or deny. This is huge time-savings, it will decrease missed findings, and should lead to better outcomes for many.

I think it’s inevitable that we’re going to see far too many companies touting AI advances of “Send 10x the email to your prospects” or “Outsource your customer support to our new AI bots”, but there is a lot of good coming too.

What are some good things you hope to see from AI in the coming years?

Unique shows for everyone

by Leave a Comment

Reading Time: 2 minutes

I recently watched the episode of “Black Mirror” on Netflix titled “Joan is Awful” and it opened up some interesting conversations.

The quick summary of the episode is that Joan comes home from work one day, fires up “Streamberry” (a spoof of Netflix) and discovers there is a new show featuring her life from that very day (starring Salma Hayek), and everyone is watching it. It’s quite embarrassing for her and leads to a variety of rough outcomes.

As the episode progresses, we find that her show was created using AI, which is how they were able to produce full episodes in near real-time. The episode gets into legal matters about using a person in a show like that (“you agreed to it in the terms of service”) and their use of Hayek’s likeness in AI, but that angle isn’t my concern for now. The more pressing matter, and one that we’ll see relatively soon, is AI-generated shows and movies that are custom-made for each of us.

Netflix and other services know a ton about us. We’re not far from the point where Netflix could have an idea for a show, but then customize it to feature things that would keep us individually more entertained. Perhaps they swap in local sports teams for each user, customize character accents, set things in homes and cars that are more appealing, etc. They could do a lot to make every one of us see a different version of the “same” show, so where does that lead?

Taking it further, they could create 100% custom programming for users. Is there a book that you wish was a movie? Done!

Engagement?

This would likely increase one kind of engagement, but decrease another. People would likely be more engaged with the super-customized shows, at least for a while, but would be less able to share thoughts with their friends because they all saw different shows.

We’re seeing that a bit already simply due to the huge volume of programming that’s available. Gone are the days of talking to folks about the show last night (Seinfeld, ER, Lost, M*A*S*H, etc) that “everyone” watched the night before. With this potential AI-generated future, we go even further in that direction.

Ultimately, no one knows. I encourage you to watch the “Jane is Awful” episode of Black Mirror if you have Netflix; it has some clearly unrealistic aspects to it, but there are pieces that just might come true before too long.

What the web is versus what we want it to be

by Leave a Comment

Reading Time: 2 minutes

In recently listening to a podcast from Gary Vaynerchuk, I realized how differently him and I see the world. While I respect what he does, and I follow a lot of his advice, it comes from a different direction.

In particular, Gary provides a ton of great ideas on how to generate increasingly large amounts of content. In the episode linked above, Gary says:

“I want to be right. I’m not talking about this because it’s how I’d like it to be. I talk about it because it’s how it is.”

When it comes to our clients at GreenMellen, I’m the same way. We work to get results regardless of how things “should be” (within the realm of morality, of course). Personally, though, I’m more interested in how the web could be versus trying to extract ever more from the web we have.

Our podcast and Meetup are both called “A Brighter Web” because that’s what we’re pushing toward, and I often share about technologies like RSS and owning your content that help bring more control back to users. It’s similar to Jay Acunzo and his idea of escaping the content hamster wheel to build more resonance with those that you reach.

Both?

There’s certainly a place for both. Jay’s approach is worthless unless you have some degree of reach so that people can see your ideas, but I think Gary goes a bit far into “50 posts per day is better than 40 posts per day”. He’s not wrong, given today’s landscape, but the ever increasing hamster wheel of content is a stressful and depressing future.

Thankfully, AI is likely going to kill that. In the short run, AI will just make the hamster wheel spin faster. Instead of 50 posts, you can do 500, or 5000. It’ll quickly become untenable for everyone, and humanity will find a place at the top.

Jay has a great two-part series where he talks about his dream for the future, and it’s a future that I’d love to see. Even better, you don’t need to wait for “the future” to start doing it; you can start right now, and you’ll likely find yourself in a great place when the hamster wheel inevitably catches on fire and burns out.

Using RSS to sniff out AI

by Leave a Comment

Reading Time: 2 minutes

While RSS has almost completely disappeared from the tech landscape, I’m still a big fan of it. I get the majority of my information via RSS, and it’s fantastic. If you’re not familiar with RSS, it’s essentially a way to subscribe to news sites and blogs. It’s not unlike getting email updates on every new story they publish, but instead of my email inbox they all go into my RSS reader and I catch up when I have time.

I’ve been using RSS for a few decades, though I’ve eased off a bit from my crazy days in the past. What’s interesting now is that it’s really easy to notice when a site stops posting human-generated content and just lets AI spit it out. I’ve seen this happen in a few ways.

First, there is a friend of mine whose blog I’ve followed for years. His posts used to be fairly short and personal, and frankly, often riddled with grammar and spelling errors. I didn’t mind, as I was just glad to hear his thoughts.

His posts had the same tone for many years, and then one day they just flipped. They instantly become 5x longer, with lots of bulleted lists on the page, a super deep vocabulary, and no typos. He absolutely switch to AI, and I was quite disappointed. I stopped following his site because I wanted to hear his thoughts, not just perfect AI stuff.

Second, I’ve seen two other marketing agencies do similar things. In their case, the tell was their pace. They went from a post or two per week up to multiple posts a day. As with the previous example, the posts become much longer, with many more bulleted lists, and vocabulary that seemed way out of context for how they normally spoke.

Sure, they can if they want

To be clear, they’re absolutely able to do this if they want. There’s no laws or restrictions against it, and I support their right to generate content however they want. Perhaps this will benefit them in terms of SEO down the road.

For me, though, I’m just sad. All three of those blogs were ones that I followed for insights and information, and now they’re just full of regurgitated fluff. I’m hopeful that this isn’t a sign of where things are going, but I fear that this is exactly the path that many more companies are going to follow.

I encourage you to fight the urge and keep some humanity in your work.

The evolution of websites

by Leave a Comment

Reading Time: 2 minutes

The internet is changing rapidly, thanks to big shifts in social media and the ongoing wave of AI. Websites have largely remained the same in recent years, staying strong as the home for every business on the internet. I don’t think that will be changing anytime soon, but the nature of the sites themselves may begin to shift.

In a recent blog post, Jeremiah Owyang had some interesting predictions for where AI could be leading, I found two to be quite interesting. The first:

As AI Agents become the dominant entities on the internet, website owners will cater to them by offering Agent APIs that instantly provide information to our AI agents, rather than simulating a human click path as AI agents do today.

In short, if we’re using AI to accomplish tasks (“ChatGPT, how do I solve this problem?”), websites will need to be able provide information to other AI bots and not just humans.

However, I think this will largely be a separate layer. Many websites use APIs today (interfaces to help tools connect together), and those will simply become more robust. The main front-end of a website can remain human-focused, and the APIs can work hard to serve AI bots.

Owyang also said:

Websites won’t go away, but they will need to evolve. When humans visit websites, the content could be AI-generated and personalized to the individual user. The era of thousands of web pages is no longer needed.

This is where the front-end of websites may shift in the coming years. The challenge will be knowing enough about each user to be able to properly customize the page to their needs. This is easy for sites where you log in (the content that you see on Facebook is clearly customized to your needs), but it’s more difficult for open sites.

While it feels like websites have tons of data on us, and some do, it’s becoming increasingly difficult to get all of that info summarized. For a website to be able to really customize the content via AI, it’d need to know a lot about you instantly, which is an idea that’s actually fading further away. This is a great thing for privacy, but could lead to some tricky developments in the coming years.

Owyang’s article is quite short, and I encourage you to give it a read. How do you see websites changing in the coming years?