Hey uguba ndubuisi, happy Sunday!
A friendly reminder that there are 55 days left in 2022. That means you still have time to write a book, train for a half marathon, or binge watch a bunch of Netflix. Whatever floats your boat.
This week, I want to talk about some fascinating developments in generative artificial intelligence. I promise, it will be fun! 🤓
I. What is Artificial Intelligence (AI)?
I'm not all that technical, so I want to establish a simple definition of artificial intelligence since I'll be referring to it often.
Artificial intelligence is a machine that takes in data, learns from that data, and uses what it learns to do things.
The ability to learn and do things is the "intelligence" part. And since it's a machine, it's "artificial."
If that sounds weird, just imagine AI as a computerized toddler. As the toddler goes through life, it learns from its observations, experiences, and the other humans it encounters. With enough time and knowledge, that toddler eventually becomes an adult who can survive and make decisions based on everything it's learned.
The reason we're talking about AI today is because it's growing out of its toddler phase and is starting to get pretty smart.
That means that AI is now capable of doing things that only humans have been able to do. Take driving a car as an example. Cars exist to help us get places more quickly, and it's not all that easy to drive them unless you're a human. Well, thanks to rapid advancements with AI, self-driving cars that drive better than humans are around the corner.
The way in which AI will transform transportation, healthcare, and other big industries is interesting, but today, I want to focus what people are calling generative artificial intelligence.
Generative AI is a specific class of machines that humans can use to create text, audio, images, and other forms of content.
Over the last year, companies like OpenAI, Midjourney, and Stable Diffusion have released AI models that allow everyday people like me to use AI for content creation.
There a many use cases for these models, but I want to focus on only two examples today since it's Sunday and you probably did not sign up to this newsletter to read a novel about artificial intelligence.
Okay, let's dive in...
II. AI Use Case 1: Images
If you've ever needed professional headshots or family photos, you know the experience kinda sucks. You pay a lot of money to hire a photographer, do your best to not be an awkward weirdo during the shoot, and hope for the best. Sometimes it works out, but even when it does, you have a "meh" experience and only a few good photos to show for it.
In fact, I went through this experience two weeks ago, and I ended up with the three photos below. They're decent, but nothing to write home about.
This week, a couple of people launched products that leverage generative artificial intelligence to help you create new profile pictures with no photographer and at a fraction of the cost.
The way it works is kind of neat. You upload ~20 photos of yourself, ranging from close-up selfies to full-body shots. Those images are given to an image-based AI engine that uses your photos to learn what you look like. The machine uses that knowledge to create new photos of "you."
Here's an example of the images I got back from Profilepicture.ai, a site created by an indiemaker (Danny) who I met in Bali earlier this year.
Pretty cool, right? And here's some more from Avatarai.me, which was created by Pieter Levels and generated $10k in 24 hours.
Of course, the photos above are some of the "best" ones. Of the hundreds of photos I received, many of them did not look like me and had odd distortions.
But even though the AI is definitely not perfect, on the whole, I was very impressed.
Some of the photos are indistinguishable from the "real me," even though they're not actually me. That's kinda spooky, but also cool.
Honestly, many of the photos are cooler, better looking, and more fun versions of me that I wouldn't have been able to dream up even if I had endless time, money, and talented designers by my side.
So for a few dollars and the click of a button, I now have new profile pictures that I can use across social media and other places.
And unlike the "meh," awkward feeling I had with my recent in-person headshot session, the experience of receiving the AI avatars was exciting. Instead of being worried that I accidentally had something in my teeth or did not have a natural smile, I was bubbling with curiosity about what the machine would pop out.
The other cool part of this experiment is that both of these products were made by individual creators who have just a laptop, some coding skills, and an eye for what people may enjoy. I used both products within 48 hours of their launch, and there have already been major improvements.
Imagine what these two creators (or millions of other people) can create over time with these new image-based AI tools?
III. AI Use Case 2: Writing
While the profile picture experiment was fun, I'm even more excited about how AI will affect the written word.
Last year, OpenAI released GPT-3, which is a text-based AI model that was trained on the entire internet and can be used to help people create written content. This week, I played with Lex, a tool that sits on top of GPT-3.
The current Lex interface looks like a Google doc. The big difference is that you can type stuff into the interface, like questions you have or unfinished pieces of writing, and the AI will help you figure it out what to do next.
For example, let's say my wife gets mad at me. That would never happen of course, but let's say it did.
Instead of Googling what to do or calling friends, I can see what Lex thinks I should do if I find myself in such a situation.
This use case is similar to how we use Google, but instead of having to sift through long articles on the topic, Lex curates what it believes to be the most helpful pieces of advice for my question.
This use case isn't all that helpful for a highly personal problem like dealing with an angry spouse, but if AI can help summarize the best available information for a specific question, I can see how it could be even more valuable than Google as it exists today.
Let's explore another use case.
Imagine I wrote a blog post about how to make better financial decisions, but I'm hungover and struggling to come up with a title for the article.
To get my creative juices flowing, I can get some help from Lex by entering something like: "Give me 10 ideas for blog posts about how to make better financial decisions."
In 5 seconds, it can give me something like this.
And if I don't like the answers, I can change the prompt or ask for more suggestions. Even if I don't end up using any of these suggestions, I'm a lot closer to having a good title with almost no effort on my end.
Okay, so those use cases are somewhat valuable, but I think the best form of Lex is having a true writing partner, someone I can collaborate with to improve my work.
Let's say I'm in the middle of writing an article about grief, and I can't get beyond the first two paragraphs. Instead of giving up or working with an editor who takes a week to get back to me, I can ask Lex to help me out.
I simply insert what I've written already, and Lex will use that writing to help me finish the article. The collaboration ends up looking something like this.
All I had to do was to write a few sentences about what the article is about, and Lex helps me fill in the gaps using a similar tone and style.
Pretty spooky, right?
What's crazy is that even in these very early days of using AI to create content, a tool like Lex is already helpful for me as a writer.
Imagine what these tools will be able to do in a decade.
It's obvious that these AI tools will have profound implications for how we write and think. And in most cases, you won't even know whether a person or an AI wrote something.
I mean, given everything I've told you, how certain are you that this newsletter was written by me and not Lex? Hard to know...
I'll have more to say on this topic in the coming months and years, including how I'm using these tools, but I wanted to give you a quick taste of what's already available as I explore.
📖 Further reading: If you're a writer in any capacity and want to learn more about the currently available AI tools, this article does a good job of giving a lay of the land: How Will We Write in 2030?
I'm also enjoying Ben's Bites, a daily newsletter about the latest in AI.
IV. AI Predictions
When you start using some of these AI tools, you're likely to experience some mix of "holy shit, this is cool" and "holy shit, this is going to replace me very soon." Here are a few of my early takes on what will happen next...
1. Generative AI is going through a hype cycle. There is a ton of excitement about these tools and money flowing into companies in the space. The excitement reminds me of what was happening in the crypto world a year ago. But as with all hype cycles, people get too excited. So while I think these AI tools are cool and have a big role in the future, they will be in the "overpromise, underdeliver" phase for a while.
2. The best thing you can do is play around. I think one of the most important skills for marketers / tech people will be knowing how to prompt AIs and use tools that sit on top of them. It's kind of like how it became really important to learn how to type or use Google as computers and the internet took over the world. Without those skills, you weren't able to compete. The same with be true with AI, and the core skillset is learning how to work with the machines. And the only real way to do that is to play around and stay curious as the tech evolves.
3. Some jobs will go. Others will stay. If your core skillset is being able to write fluffy listicle blog posts or to generate mediocre headlines for social media ads, then you will likely be replaced by AI. The tools are already good enough to replace this type of work. But while AI may replace something like entry level writing, I think it's a very long way away from competing with someone who lives a fascinating life and can write a well-crafted story about the unique perspective they have from living such a life. That's good news for memoirists, but not for social media writers.
4. Even if some jobs go, others will pop up. From the advent of the printing press to bicycles to computers and smartphones, every new technology has been met with resistance by people who don't want the world to change. Steph and I talk about this dynamic in our Technology People Feared pod. But time and time again, history has proven that even when new technologies come in and replace current jobs, a new set of jobs opens up. And to get those jobs, you often need to re-skill and stay adaptive in a fast-changing world. So while I think AI will replace some jobs, new ones – including those we can't imagine – will pop up. Just think, 20 years ago, there was no such thing as a "social media manager." But now you can make six figures doing that job. AI will create similar opportunities.
💰 Money x Relationships
Steph and I published a new episode on the podcast about Money and Relationships. We give an inside look at the taboo topic of how we think about and manage all things money.
You can listen to the episode here.
As a reminder, if you have any topics that you'd like us to talk about, feel free to reply to this email to let me know.
Thanks for tuning in, and see you in two weeks.
Cheers,
Cal
