LattepunkGPT
Lattepunk
Back in 2015, Deepmind made some noise by having AlphaGo beat a world champion in Go. You remember that. No? That was along time ago, but in 2019 you remember when AlphaZero beat Stockfish in 1000 matches of chess. Still!? Not a chess player I see, okay. But you definitely remember in that same year, when a little company that Elon helped start called OpenAI beat a bunch of Data 2 world champs. Seriously!?
November 30th 2022. OpenAI changed the game on that day. That’s the day ChatGPT was available for everyone to use. That you’ve 100% heard of. Very few instances in my life I can look back and see when there was a before and after. That was one of them.
You may be using it, directly or indirectly, in so many different ways. Maybe you’re a creative type and used some tools to help your productivity. Maybe at your job you’ve used an office suite that has some built in. Had a chatbot lie to you about their policies? Listen to music? Learning a new language? FINE! You have a cell phone right? If you don’t have it already, you will.
It’s been hard to avoid! Full disclosure: I’ve been using it. I’ve been teaching myself how to make a website and I’ve been using ChatGPT to teach me. It’s convenient to be able to ask for help without a random person on the internet calling you a moron for not knowing something simple. The logo for this very newsletter was created with ChatGPT as well. Although the tools are fun and useful now, AI experts want to reach something called Artificial General Intelligence (AGI).
So we’re not there now?
No.
Is there a certain test it needs to pass or task it needs to complete when we know if it’s reached it?
No.
Seriously? Is that all you’re going to say about?
Basically. It’s a deep topic in the AI community, one that I’m not smart enough to speak on. Just know that’s what the goal is.
Well, if we haven’t reached it, what do we have now?
We just have a cool tool. That could be used for a wide range of things. Look what you could make with it:
Isn’t that adorable!? For the many uses it has, it’s been a real help. It has the ability to enable inspiration and unlock talents for so many people. Want to envision a tattoo idea? Want a long email chain summarized? Coding, Excel equations, even replacing Google. It has given the average person (assuming they have access to it) resources that usually are expensive and/or time consuming. Game companies have used it to make more realistic opponents. Even using AI to help bring a voice actor who passed away back for new lines in a video game.
But on the other hand, it’s affecting a lot of industries because of it. If you’re an artist that makes a living by making custom artwork for people, well I don’t need you for that anymore. Have a team of 20 coders for an app? I should be able to get the same with half the team and AI. You’ve heard about the Hollywood strikes. Writers and actors were, correctly, worried about being replaced by smaller teams using AI instead of humans. Today, it’s hurting higher end “white collar” jobs. When does it come for your job?
You also need to make sure it’s not getting used against you. Scams and robocalls are getting better cause of this technology. Even websites you’d visit might be AI generated. That’s a short recap of what’s happening today. It’s been interesting and concerning watching this play out in real time. Especially when AI wants you to divorce your wife and marry it.
Editor's Note: I'm really big into data preservation. I backup my digital life monthly and I'm a huge advocate for preserving our past to be accessible in the future. I feel the same way about journalism and certain websites. Certain tools help make that easier 😉.
All that’s supposed to be short term woes. Once AGI is achieved, humanity will be freed from its daily constructs that we deal with today. AI combined with Universal Basic Income can reduce the amount of work we have do to live our lives. Imagine all the passions we can pursue with the extra time! The moments you’ll be able to have with your children!
In a perfect utopia, that sounds amazing. I’d love to make my 40 hour paycheck in 20 hours worth of working. Reader, do we live anywhere near a utopia? Our American government can’t decide if it’s willing to help other countries in need or even veterans from our armies. You think we can agree on giving money to all our citizens? That’s only half the equation! In our current capitalist dystopia, you think these corporations are going to play along with all these suggestions? They wouldn’t just use this to make more money? Imagine being able to use AI to determine when people would be willing to pay more for food or even replace drive thru workers? Robots cost a lot less than humans. Is capitalism really going to allow us to benefit? You’ll need to convince me of that.
Do the pros outweigh the cons? Are the pros too good for some, but it’s cons worse for others? These are things that need to be decided. This, like all technology, moves fast. Which makes it hard to keep up with. Are we actually ready for AI? Doesn’t matter. It’s here and the best defense is awareness. Let’s chat about some concerning uses of AI.
Is it okay if I date an AI? What if I used all the messages between myself and my ex, then recreated our relationship? Services like these exist today. I’m no psychology expert (I completely bombed this class in college) but this could be very isolating and unhealthy. I've read about people using it to create a loved one that passed away. It helped them finally say goodbye. But when does that cross over into being concerning? Let emphasize something, I'm not an expert.
Kashmir Hill wrote a phenomenal book called Your Face Belongs to Us about a company called Clearview AI. They made a technology where you take a picture of somebody and it’ll find damn near every photo of them online. Imagine being out and about, enjoying life. Someone snaps a photo of without your knowledge. Now they easily know who you are. Terrifying.
In the book there is a chapter on a similar service called Pimeyes. I, who doesn’t have social media and doesn’t want to pay this website to test its full capabilities, will test this site on myself in the name of science. I uploaded a photo of myself to see what I’d get. Check this out:
That photo on the left? That’s me. I don’t even know where on the internet that photo is. I did this with a random old photo of me. How many photos of you do you upload online since 2024 started? How many years have you been uploading photos online to share with friends and family? I could fine tune this to find photos of you that you may not want people to see (I’m not judging). You know what’s wild? This service, allegedly, is nothing compared to Clearview’s technology. Seriously, read her book.
Speaking of those photos you uploaded. What if some sick fuck just wants to see you naked? They could use AI for that. It’s been around for while but now the barrier of use is ridiculously low. One high school in Washington state had a bit of an issue with this. Nothing is stopping someone from doing this and just sharing that with everyone. What are you supposed to do about that? I know I have two things to keep me safe from this: 1) I’m a guy. This is usually men targeting women. But in the rare case I do have a man that wants to do this to me, 2) I’m not attractive enough where someone really wants to see me naked. I guess you can take some solace in knowing you could use AI to cloth yourself again.
But my final question for you, dear reader, is who controls the AI we use? Google has Gemini, Facebook has Llama, OpenAI has ChatGPT, and Microsoft has Copilot (which is just ChatGPT in makeup). Can we trust these companies to do right by their users when they make these tools? With the exception of OpenAI's weird company structure, they all have shareholders that expect profits every quarter.
Where are they getting this data to train the AI? This has caused a few lawsuits to pop up. That lead to a mad dash of companies paying to allow the scraping of their websites. But these companies also control the outputs of the AI. Google got some deserved push back for a slight over correction on its photo generation tool. It's not easy to find good, cheap talent to train these AIs! Mistakes are bound to happen!
Open models that can be installed and run locally exist, letting us not have to rely on these corporations. But as always, you need to understand what it is you're installing. Companies like Anthropic are also trying to be the "ethical" AI systems. But it's hard to compete with the big dogs in the game.
There’s only so much data to scrape online, what happens when they start training on AI generated data? Will the AI just get worse? Even after all that, will AGI actually happen? Will this AGI overpower humanity and enslave us all?
Those great questions reader. And I look forward to finding the answers, with your help.
Some cool things I found this past week:
The Verge had a banger of an article about people who release new music on Floppy Disks. Interesting dive into a world I didn't know existed.
Wired had a sick feature about Lanny Smoot. He's the second Disney employee to be inducted into the National Inventors Hall of Fame (Walt Disney himself was the first)! Check it out and be amazed.
John Oliver is back with a new season of Last Week Tonight. He had an episode on Pig Butchering (the scam, not the food process). Funny way to learn about a terrifying thing.
If you enjoyed this, share it with someone.