AI versus Marshy #45: the death of the Internet, private knowledge AI, and pitching spark


Hello and welcome to another edition of AI versus Marshy!

This is the newsletter that demystifies AI and gives you things you can think about and action TODAY.

Straight-up, I have TWO apologies I want to dish out.

  1. Last week I tried uploading images into a new backend to make the uploading images process easier in ConvertKit. This looked fine in preview, but I skipped test email, and some of you got an email with no images. 2/3 of you got a second email with the images fixed. My bad!
  2. Running later this week with challenges in child-rearing. It was a tough week!

Okay, back to normal programming.

Lots of things are happening at the moment.

Overall literacy about what can and can’t be done is increasing.

Instead of the tech now providing the omgwtfbreakthrough mind melts that ChatGPT’s launch delivered, we’re seeing use cases, tutorials, and more content of people actually doing useful things with AI.

This week covers:

  • More philosophical waxing about the death of the Internet as we know it
  • Building a personal RAG - and what that is
  • I pitched ADHD Goal Machine

I’m keen to get stuck in, so let’s start writing the glue and get sticky with it.

(Too much?)

-Marshy 💪

The Internet as we know it is dying

Via Where’s your Ed at?.

One of the nerdy things I like to do is look back on old articles (I have a lot of them saved) and cross-check predictions that were made.

For example, reputable tech writer covered Magic Leap in 2016.

It covers a “photonic light chip” that will fundamentally alter the world of virtual/augmented reality.

Since then it’s gone bust, Apple has released the Vision Pro, and I covered Google’s build of a new demux last week - a superior photon semiconductor chip that was built with AI.

More recently Edward Zitron writes about the Internet as we know it dying.

It was written in March 2024, and he shares this pearl early on:

It’s now May 2024 and here’s the Reddit stock price in USD:

The rest of Edward’s article talks about the death of the Internet. He lambasts SEO, “handmade” goods on Etsy, and LinkedIn corpo “shit-posting” (my words) using AI tools.

All of these things are happening.

I really like his point about Google Search quality declining too.

I don’t have data - I just know in my bones that I could find interesting articles on Google with search 10 years ago - and now that’s impossible.

Every single top article is optimised into nothingness - I read it and I get the information I was after - but never learn anything.

He argues generative AI is making this problem worse.

I disagree.

Generative AI is increasing a problem that was already there.

This is the way Internet and tech companies make money:
1. Create something useful or entertaining
2. Achieve a critical mass of users
3. Profit off them and protect your moat

Examples:

Facebook - gather people’s content and put ads against it

Google - gather people’s websites and get people to pay for ads to send people to it

OpenAI - gather vast multitudes of data and train an algorithm and get people to pay to use it

The gathering is different for each company, but once the vast amounts of funding and/or profits are gained, spend a portion of that war chest on protecting, refining, and improving your advantage.

Generative AI is providing new ways to protect the advantage - but my argument is why does the Internet need companies that are pushing advantage?

It’s a shittier experience and that’s what we’re seeing.

A more thoughtful take on this was written by Maggie Appleton - The Expanding Dark Forest and Generative AI (I’ve linked this before).

Both articles lament the loss of the Internet as we know it.

I think we can move towards something more community-led, and collectively better off for everyone - but the Internet will probably get worse before that happens (or we’ll all keep having better interactions in our private Slack, chat, and WhatsApp groups).

Science-fiction takes these different stages of Internet to extremes.

In the Cyberpunk universe, the "blackwall” is a state-sanctioned AI that prevents rogue AI from the “Old Internet” getting into the New Internet because it’s become so dangerous.

Building a personal knowledge AI

Via AI Jason.

An idea I’ve held for a very long time is having your own personal pensieve.

In Harry Potter, this is a magic item that lets you store memories away to return to later.

The world of “personal knowledge management” is a deep rabbit hole - with the core of it being a system for managing your online life (articles, thoughts, ideas, snippets, etc.).

There are a lot of thought leaders in this space out there now, Thomas Frank is one who loves Notion, and Tiago Forte is another who talks about having a Second Brain.

I use some PKM (Personal Knowledge Management) principles myself.

I’m yet to find a good AI solution for this, and then I saw this video explaining why:

It’s a longer video. The presenter does a great job of explaining the slide above, and then shows a slide after explaining it deeper 🤯

Essentially an AI version of personal knowledge management is a RAG (retrieval augmented generation).

The video explains how to build one with the open source LLM Llama 3 by Meta.

If you want a look at how good/bad it is - Australians can now interact with it within their socials.

The video prompted a few ideas for me to think about:

  • The format of talking head and slides is one I can do until the cows come home and am thinking I might trial some videos this way
  • Using the video to build my own private RAG - is it hard?
  • Some more thoughts about this newsletter and how to separate each article into a headless CRM in Airtable

This stuff is still pointy-nerd end but for a more practical and approachable way to use AI today - I recommend watching/listening to this video/podcast:

Dan talks about using GPT’s very practically as a method for helping him and his team craft words and it might inspire some of you!

I did a live pitch about ADHD Goal Machine

Via me.

I joined a coworking space a few weeks ago at Melbourne Innovation Centre and participated at the monthly networking event Suburban Spark.

It was a good excuse to talk about ADHD Goal Machine out loud and get some live crowd feedback.

I’ve got a few ways I can move this forward but I feel like it will be a slow burn - I simply don’t have the time/energy to go any faster at it.

Here’s a write-up for the event and snap below.

That’s it for this week - a bit of a longer and deeper one that I enjoyed unpacking.

The only consistent thing in tech/AI world is change.

A few interesting moves and projects on the horizon for me - I feel very lucky and grateful to have a career that is dynamic and can help many different people, industries, and areas.

Thanks for reading, and remember we’ve got this!

-Marshy 💪

p.s. I enjoyed reading through this article about “wellbeing proxies”. I have like 300 of these. An example is when I notice I’m in a shitty mood for longer than is usual, it almost always means I haven’t run in a few days. Easy fix! ✅

Early Growth School

Continuous tips, lessons, advice, and direction on growth for your early-stage startup by a growth marketing pro.

Read more from Early Growth School

the results are in. Last week I said I had a bit of spare time and wanted to test more video with shorts. This is what happened in the last week: Not bad right? Well actually in the scheme things - it’s unprecedented. Here’s the lifetime view count of my channel: The number’s pathetic. It’s never been a focus, and while I love presenting, have never been inclined to go hard at it because I have other things to do (like work for clients, raise my family, get sober and completely transform my...

short example

a tierlist for growth channels. Have you heard of tier lists? They’re very common in gaming world. Influencers like to put together lists ranking characters, items, or selections and opine on them. I had a quieter week as I work through a project exension and thought: Do you know what the world needs? A tierlist for growth channels for startups. But not just any tierlist, but grabbing and ranking the channels from the book Traction. Written in 2015 Here’s the channels from the book: •...

doing the work. “So you’ve done the work…” I met Mel Greblo through my work in the impact space. She does a number of amazing things, including founding Banksia Academy. A destination for helping women create their own employment pathways to help remove themselves and recover from domestic violence. She now leads Humans of Purpose. “Oh yes, I have.” If you read into the question literally - you can ask things like: what work? in what context? is it this or that? what kind of work? But if...