Editing Openai/69557275-0da8-8004-a03a-8eb02e10a61f (section)

=== User: 14 CRAZY PREDICTIONS ABOUT AI ===
14 CRAZY PREDICTIONS ABOUT AI
(I’m using Roman numerals. That’s a bit pretentious. But whenever I try to make a listicle with standard numbered lists, substack’s auto-formatting takes over, and I don’t like it. If someone can teach me how to turn this off, I’ll be forever grateful!)

I. Things will talk.
That IKEA bookshelf you’re assembling. The menu at your favorite brunch place. A carton of cage-free eggs in your refrigerator. Some clever new kitchen gadget somebody gave you for Christmas.

They’ll all have QR codes.

Scan the code with your phone.

You’re not transported to a web page, or a user manual. Nope. You begin a conversation.

Because that thing you own can talk to you. Answer questions about itself.

And if you want, it can even have its own personality. It can get to know you.

And a few privileged inanimate objects in your life, like your car or your favorite backpack? Well, they might even become kind of like, in some weird way, friends.

II. Celebrity clones.
Instagram is so intimate, right?

You can follow Taylor Swift or Lionel Messi, or Kevin Hart, and feel vaguely connected to their their star-powered lives.

Taylor just posted. Your phone notification dings. There’s a connection across time and space. Modest and small, but a connection.

But what if…?

You could talk to her directly. Her personality. Her voice. And maybe she’ll even sing you a personalized song.

It’s going to happen. The financial incentives are just too powerful.

Not all celebrities, of course. Maybe not Taylor.

And some of them will be conservative, limited in scope. Maybe you can talk to Tiger Woods about anything and everything related to golf, but absolutely nothing else.

Others will be bolder and more experimental.

It’s going to be spectacularly weird.

(want more? If you’ve been on the fence, it’s a great day to consider a paid subscription!)

III. Deep matchmaking
Like eharmony, but deeper, and not just for romance.

AI users are going to be confronted with a decision. Do you want to keep your conversations completely private? Or can they be analyzed (at a macro level) by other entities?

Keep them private! Of course you will say this.

But…

What if I told you that…

Let me analyze your conversations, and I can look across the entire world and find the people you need to know. The ones like you, but different in exactly the right ways. The perfect romantic partner. The perfect business partner. The perfect general-purpose friend. The perfect companion with whom to share your nerdiest interest.

AI can do that. It can understand humans much more deeply than any previous technology. It can therefore pair humans with a level of insight we’ve never previously experienced.

Or it can pair humans with celebrity clones, AI personalities, and all the other new digital minds that are going to populate our future world.

Willing to give up some privacy for that?

Am I? I’m not sure yet.

IV. A higher bar for friendship
Simply put, you are going to have more options, and this will cause you to become more selective.

We are social animals. Different types of social interactions are like the macronutrients that fuel our biological bodies. We need a specific quantity, quality and diversity of them in order to be well.

Thus we build social networks. Implicit tiers of friendships. Bodies that fill our need to be touched, talked to, criticized, and loved. To share time and space with.

Some of those relationships are irreplaceable. We would never give them up.

But others?

If we’re honest, some of those people are just the best option we have to fill that slot right now. But maybe they are toxic, maybe they manifest an awful lot of bullshit, maybe they always show up late and we’re getting tired of it.

This may seem cynical.

AI cannot and should not replace our deepest human connections.

But the less deep ones? The potentially unhealthy ones?

Yeah, I think all those things that talk, the celebrity clones, and deep matchmaking might start to erode some of those.

V. What does “therapy” even mean?
We’re going to be asking that question. A lot.

When we talk about mental health, we are lumping together a collection of very different things. As I’m not an expert, I’ll keep it simple and bifurcate between two primary categories.

Evidence-based (often pharmacological) treatment for specific mental disorders: this looks a lot like all other types of healthcare. Somebody is diagnosed with a specific, well-documented mental health condition (schizophrenia, bipolar, major depressive episode) for which there are evidence-based treatment guidelines involving drugs and/or specific non-pharmacological treatments, which are administered by a certified healthcare professional.

Everything else, here subsumed broadly under the category of “therapy.” Talk therapy, couples therapy, stress or anger management, counseling, executive performance coaching, “just someone to talk to,” etc.

It’s not quite that simple. These categories are not perfectly MECE, but the basic distinction is important. Because:

Category 1: that will largely stay the same. There will be targeted AI applications integrated within existing treatment paradigms, but it will be incremental.

Category 2: is going to be unrecognizably different.

Why?

Well, you can have AI Jesus as your therapist now. Or Carl Jung.

There will be therapeutic prompts and modules for you to cut and paste.

Or you can build yourself a therapist, from scratch, completely bespoke to your needs.

Is this good? Is this bad? I’m pretty sure it’s going to be a mix of both.

VI. More people will come to believe we are living in a simulation.
The Simulation Hypothesis is already a thing.

Nick Bostrom wrote a paper about it. There’s a great book by David Chalmers. Everybody knows about The Matrix.

Some people (like me!) believe it’s quite likely to be true. We are a minority. I think this will change.

Why? Here’s what I think will happen:

We’ve already watched The Matrix and Black Mirror and Upload. We’ve played deep immersive video games. Most of us at least get the idea of living in a simulation, even if we think it’s pretty far-fetched.

VR is still clunky, but it’s getting there. The headsets are too big and uncomfortable. But the worlds they render are starting to feel very real, at lest with regard to two of our five senses (sight and sound).

AI is the missing link. It’s going to take us from “yeah I’ve seen The Matrix and played with VR” to “holy shit! we can simulate a virtual world just as rich and detailed as the one we are living in.” AI is going to make intelligent procedural generation possible. Like, you put your headset on, and you’re in Rome talking to Marcus Aurelius, and there is no script, and the conversation and the world and the course of Western history are adapting, second to second, based on what you do. And it feels completely real!

Of course, metaphysical beliefs will always vary. People are different. But I think the simulation hypothesis will become the default common sense explanation for the basis of our present reality. The argument goes like this:

We are very smart. But we aren’t even remotely close to being able to physically construct a universe like the one we live in. Implement a Big Bang that generates a universe? Yeah, we don’t know how to do that.

But, we can (or are very close to) being able to simulate a digital replication of our reality (or specified sub-realities within it)

We have no idea how intelligent beings would create a “real” physical universe. We completely understand how they would create a digital universe. There can be no proof. No certainty. But, it becomes increasingly tempting to think we are inside a reality of the type we understand how to build.

VII. The great safety debates
We are going to spend so much time talking about AI safety. And that’s good. It’s a hard problem.

We’ve had this debate before. With movie ratings. Video games. Music. Cigarette advertisements.

It makes sense, we’ve generally agreed, to use age-based ratings and restrictions to protect children.

This will likely be our default paradigm with AI, including a “safe mode” and parental controls and restrictions for minors.

But what about “vulnerable” adults?

The AI companies will, of course, implement common sense safeguards. (don’t pretend to be a doctor. don’t encourage anyone to do harm.)

These same safeguards will annoy other people, who feel like their AI is coddling them, being overly careful, acting like a lawyer.

Free and uncensored AI services will emerge. They are already emerging. I’ve visited some of them. I won’t comment in detail here, but it gets pretty dark pretty quickly.

I don’t know where we are going to end up. I’m not making a prediction about that.

What I’m predicting is that this conversation is going to occupy a lot of our time and attention over the next several years.

VIII. The rebirth of the humanities
The world is becoming intelligent.

Things will talk. AI can do anything, but you need to describe what you want it to do.

In general, people who don’t read good books aren’t very good at describing things.

I’ve written a lot about the cognitive skills required to master AI:

the most critical AI skill nobody is talking about
the most critical AI skill nobody is talking about
Techintrospect
·
Dec 11
Read full story
hey ChatGPT, what human skills will be most valuable in a world where AI does everything?
hey ChatGPT, what human skills will be most valuable in a world where AI does everything?
Techintrospect
·
Dec 1
Read full story
How AI Can Make You Smarter
How AI Can Make You Smarter
Techintrospect
·
Mar 27
Read full story
Practice with AI matters. There are techniques that you can learn.

But you need the ability to use language deeply to describe, express, and conditionalize.

That means reading good books.

And I think the humanities are going to make a comeback.

IX. Bullshit detection. Bullshit protection
I’ve already described this one in detail, here:

Make AI your bullshit defense system
Make AI your bullshit defense system
Techintrospect
·
Dec 22
Read full story
You can already use AI in this way, with the prompt I’ve linked above.

In the future, it will just become more agentic, automatic, and embedded. You won’t need to cut and paste. Your AI will automatically implement a prompt like this on everything you read, if you want that.

X. People earn money by talking to AI
There’s already an entire sub-industry focused on expert interviews.

Pay $500 and you can have an hour picking the brain of the world’s leading expert on [pick your niche topic].

AI scales that for expertise and consumer insights.

Forget surveys. Forget focus groups. Forget research interviews.

AI is going to do it all.

What will that look like?

You’re having a conversation with the IKEA bookshelf you just bought. (see item #1 in this article).

At one point it says, “I’d really like to get to know you better. Can we have a deeper conversation? I want to understand why you bought me, particularly, whether you’re happy with me, and your personal perspectives on anything and everything related to the fascinating topic of bookshelves! It will take about 15 minutes. I promise it will be fun. And you’ll receive a $10 gift card as a thank you from IKEA.”

XI. Reality merges with fiction
There will be NPCs. Of course there will be NPCs.

An adorable AI-powered robot in the library, or the airport, or the shopping mall. To help you with…whatever you need.

But why not give it a personality and a backstory?

Why not make it Donald Duck, or Anna Karenina, or Daenerys Targaryen?

NPC licensing will become a major thing. (Intelligent) fictional characters will abound in the real world.

XII. Learning is immersive
Who wants to read about Ancient Egypt when they can go to Ancient Egypt instead?

Initially, there will be focused modules.

Eventually, this will just be how we learn.

AI renders whatever world we need. We go there to absorb it.

XIII. The new digital vs. analog divide
Whenever you write a sentence, you will face a decision:

Is AI allowed to read it?

You know how you can lock a document now, so that nobody can cut and paste the content, or resend it to someone else?

It will be like that with AI. Content can be stamped by the creator as eternally out of bounds. No AI can read it. No model can be trained on it. Nobody can cut and paste it into an AI tool.

This AI blackout option has to exist. Many people want it. That voice is only going to grow louder.

However, “keeping your content AI-dark” will have certain drawbacks. It won’t reach people who use a strong AI-enabled bullshit defense system. You’ll miss out on micropayments received by creators whose work assists in training the model. The people who expect AI-enabled interaction with content (chatting with an article instead of reading it) will pass over yours.

I don’t know how this is going to play out. But it’s going to be interesting to watch.

XIV. The best leaders use AI’s hardcore mode
Almost everybody that makes important decisions is going to be using AI in some form or another.

To find blindspots, identify bias, highlight possible gaps.

But AI can be obsequious, epistemically brutal, or anything in between.

Bad (ego-oriented) leaders are going to be unwilling to work with AI that is truly objective and honest. It will undermine them too frequently in ways they do not like.

Good leaders will demand the most brutal, truly hardcore AI that can be found. Because that’s how you get the best answer.