The Road Ahead
What's next and what does it have to do with me?
In this edition:
My thoughts
Notable headlines
Learning, Tools, and Experiments
AI MISTAKES
My Thoughts
MISTAKES HAPPEN. They are learning opportunities.
Life is messy. We often don’t have the full context.
It’s not uncommon for us to glaze over details and ask for things we don’t fully understand.
Is this human existence? AI chat sessions? Our relationship with other people?
YES.
But there is a powerful tool that helps to correct this. When we make mistakes, if we can get the right feedback, it becomes a valuable opportunity for improvement.
So what does feedback look like today? How can we do better? And what does that have to do with the latest and greatest in AI models, tools and techniques?
Over the last month, we’ve seen some incredible progress with open models, a new country has taken the scoreboard by storm, and political organizations (governments and others) are taking it upon themselves to regulate. The best models in the world keep getting better and there seems to be new capabilities released on a regular basis.
Take this generated image from above. There is no place in the world that looks like that. It does remind me a bit of a place I visited last summer in Scotland, but it’s decidedly not real.
The lack of reality is the problem. Whether it’s the feedback, the guidelines or even the understanding of what’s going on, as a society we have to keep learning.
Mistakes are being made. Some are human. Some are machines, and some are just plain lack of attention or intention.
So, where do we go from here?
My personal position is that the box has been opened. You can’t put it back in. We have to deal with what we’ve created for better or for worse.
I'm actually a fan of giving this technology to the world and setting it loose. I’m actually an advocate of going one step further. The way I see it, it’s on all of us to find the GOOD ways we can use this technology and put it to use, if only so that it drowns out the bad.
I keep seeing people say, “Oh, we can't trust anything on the Internet now because of AI!”
I can’t be the only one who thinks this, but I’m certain that ship sailed many years ago. Anyone with a blog can say anything at any time.
It is the automation and scale that make all the difference now, and that’s available to everyone.
So, let’s build.
Let’s deploy.
Let’s create!
Let’s find ways to overwhelm the negative with the positive. This technology has already seen incredible availability, applicability, and democratization.
I'm actually writing this newsletter because AI helps me make it happen.
The ball is in our court. Instead of letting someone else decide how it gets used, we can activate it today and advocate for what’s good and right in the world like never before. The sheer power of artificial intelligence coupled with automation can be used to address real problems today, and I know that we can overwhelm the bad.
It's actually the ultimate expression of freedom.
Many people have more to say than they admit, but fear overcomes their actions.
What can you achieve if you set your mind to it? Anything!
And now the reasons you have for not starting are eliminated with the power of AI.
You can create. You can build your dreams. You only need to start and be willing to start, knowing there will be MISTAKES.
So, let’s use all this great stuff to make the world a better place. Let’s focus on the good things happening and remember that iteration and feedback are key to improvement.
Now, let’s go learn, make MISTAKES, and build a better tomorrow.
Notable Headlines
The models keep coming. The models keep getting better.
AI Gadgets aren’t getting great reviews, but it’s only a matter of time before there’s a hit (maybe Apple?). Science fiction has already pointed us in the right direction and provided inspiration, but implementing great tech is hard.
Learning, Tools, and Experiments
OpenAI
OpenAI ChatGPT offers more features to free users. If you haven’t created an account with OpenAI because you’ve been avoiding paying a monthly fee, now is the time to do so.
Watch OpenAI's Spring Update Event
The announcement this week also means that anyone can start using GPTs as listed in the store.
But wait, there’s more! So much more. Check out Friday’s Live video covering all the big news items in the release.
And, while it’s not the kind of learning we most like, Once again Sam Altman find’s himself with some unwanted attention. In this case, it’s the voice ‘Sky’ in ChatGPT. It’s a little too familiar for some tastes. To make matters worse, Sam made mention of “HER” and that hasn’t helped.
Will this be a MISTAKE? Only time will tell. Meanwhile, it may be an opportunity for for learning.
If you haven’t seen the movie Her, it may be a good time to catch yet another movie about AI.
I hear you asking already: So what are the essential AI movies?
Here’s my quick rundown of the essentials.
Her
The Matrix
I, Robot
Ex Machina
2001: A Space Odyssey
Terminator
Hollywood tends to choose the most violent examples, while I prefer the more helpful variety, so here’s a few science fiction books that give a deeper perspective:
Ender’s Game + Speaker for the Dead + Xenocide (Ender’s Game is actually my most favorite Sci-fi stories, so let me hear from you if you read and enjoy it)
The Moon is a Harsh Mistress
Asimov’s Short Story Robot Dreams (I would say Asimov was the first to imagine AI in his stories, but it’s goes beyond Robot Dreams)
Asimov’s Robot book series
Caves of Steel > The Naked Sun > The Robots of Dawn and Robots and Empire (which follows the Foundation series, also recommended)
There’s plenty more AI learning in books, but I’ll call it a day with just the science fiction for now.
ElevenLabs expands into Music
Here’s a preview of ElevenLabs Music. Generated music, no edits. I’ve been playing with ElevenLabs for a few months since a colleague shared an audio snippet with me that replicated his voice, but it wasn’t him. It’s incredible what’s possible, and ElevenLabs is doing some great work. This music is impressive, but there’s still work to be done.
Checkout ElevenLabs and get started for free!
Speaking of what’s possible, checkout this “Reid Hoffman meets his AI twin - Full” video on X. Reid uses Hour One, ElevenLabs and a custom GPT (OpenAI) to create an AI twin.
Personal-AI is an assistant that works using Apple Shortcuts. If you have an iPhone, you don’t need to wait for Apple to release a better version of Siri, you can use it with Llama 3 today (the bad news… it’s still requires an internet connection).
Google Cloud Experiments
Google released a huge amount of information at Google IO this week, and I can’t possible cover it all. The truth is, for much of what they released, the value is in getting your hands dirty and playing with code and learning hands on.
Your feedback is valuable and I’d love to be able to connect with my readers.
Don’t be shy, I’m just a normal guy, learning about AI.
MISTAKES with AI
It wouldn’t be an AI MISTAKES newsletter without discussing some element of the framework, getting started, or pointing out the places where things didn’t quite go as planned.
This month, I’d like to discuss the biggest risk I see with AI.
Many are confounded and, quite honestly, scared of AGI. But before AI can exceed our human capacities, we will have to survive when people assume the AI is better and don’t bother to check things out for themselves.
In September of 2023, a team of researchers released a working paper: Navigating the Jagged Technological Frontier: Feild Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality.
I have discussed this paper multiple times, and this likely will not be the last. Unfortunately, as a society, we will have to learn this lesson over and over again, so it bears repeating for this audience.
To quickly summarize, it covers several important dimensions and shows valuable insight into how AI changes our abilities (both good and bad).
Taken from this report is the below graphic.
The most important part of the chart is that little read portion at the bottom left corner.
This data point represents what happens when humans stop thinking and only rely on AI.
To take directly from the paper, there are three critical ways this happens:
Overreliance on AI: When individuals depend entirely on AI, they may become less likely to critically evaluate the AI’s output. This can lead to a decrease in performance for tasks that are outside the AI’s capabilities.
Blind Adoption: Users accepting AI-generated solutions without scrutiny. This behavior can result in bad results, especially for tasks that AI cannot handle.
Unengaged Interaction: Professionals who do not engage deeply with AI and fail to question its outputs tend to adopt AI suggestions without verification, leading to potential errors in judgment or decision-making.
In summary, Too much Trust and a Lack of Validation.
While automation of a person's work with the aid of AI is really valuable, if we stop thinking, we will have a problem. We’ve already learned this lesson repeatedly with tech like autopilot capabilities in aviation.
Don’t make this mistake. Keep your guard up. Use a critical eye. It’s easy to let your guard down and let it slip through. But with the right attention and AI at our side, we can build tomorrow better than today.
As I close out this edition, let’s make the GROWTH AI MISTAKES, not the silly, lazy, disastrous mistakes we’ve discussed this month. Go, try new things, start getting your hands dirty, and discover how this technology can help us grow and improve.
Go make some AI MISTAKES.