The Law Is Not Looking Good for AI

Influencer makes $70k in a week with AI + Google announces announcements

Welcome to edition #15 of the “No Longer a Nincompoop with Nofil” newsletter. I have a very special announcement to make, the first step in my vision for the future of AI.

I’m starting an AI Dev company 🚀 

A few businesses have reached out to me asking how AI can help them with their products. Naturally I simply don’t have time to build products for free, so I’m going to start providing AI services. I’m already helping a startup train an AI bot on all their documentation to help provide support for their in-house team as well their customers.

I started this newsletter to help people stay informed on what is happening with AI. It’s impact can’t be overstated, particularly on work.

If you’d like to use AI for your business, feel free to reply to this email or email me at [email protected]

Here’s the tea ☕

  • AI & Law ⚖️

  • Google playing catch-up 🏃‍♂️

  • AI’s cure to loneliness 💊

The complicated mess that is law

Former Google CEO Eric Schmidt did an interview in which he says guardrails are absolutely necessary within AI, but governments should not be the one’s to create them. He doesn’t think they can get it right because he thinks “there’s no way a non-industry person can understand what is possible.. no one in government who can get it right”. It’s a tricky situation. I want to outright agree but it’s not so simple. Let me show you why.

The EU is planning to do something incredibly stupid. In fact, I almost want to call it corruption. The AI Act has a number of worrying aspects, one of which is the idea that any model available in the EU has to go through an extensive licensing process or otherwise be fined up to £20 million. I’m not only talking about models released by Google or OpenAI. I’m talking about open source models - so if someone playing around with an AI model in their garage and uploads it and it’s available in Europe, not only are they liable, but GitHub, the platform that hosts essentially the entire worlds code is also liable.

But the most ridiculous and downright crazy aspect of the proposal is it’s regulation on API’s. API’s allow practically anyone to run and implement the functionality of an AI model without having to run it on their own hardware. This is essentially how most people have been building AI powered tools or even systems to help build tools like Langchain.

Under this proposal, if someone using an API get’s the model to do something new, they would have to get this new functionality licensed and certified. This is absurd. Not just in the logical sense but what it’s essentially doing is stopping every single individual, entrepreneur or startup building with AI. The only winners from this nonsensical proposal is corporations. There are many other details which are so silly it’s hard to imagine this might actually go to EU parliament. If this actually goes through, we will almost certainly have a monopoly on AI models and innovation. The age of AI innovation would have lasted less than a year in the EU before it was killed.

Clearly governments have no idea what they’re doing. They’re going against what the community wants and is advocating for, and ultimately playing into the hands of big tech. The problem here is that governments do need to be involved, but not like this. Unfortunately I don’t see how this plays out well for us. We can’t trust companies to self regulate and we can’t trust governments to do it either (or anything else for that matter). So we’re in a peculiar predicament, one in which I think we lose no matter what happens. The question is how bad it will be. I suspect it will be awful for the average folk like you and me, kind of how it’s always been.

Google dances

Google announced a tonne of things last week, but again, they are just announcements. They announced Bard doesn’t have a waitlist anymore and is available in over 100 countries, but most of Europe doesn’t have access to it. They announced PaLM 2, the next iteration of their LLM’s that will power Bard, but we have no idea when it will be out. They announced AI across Photos, Gmail, Docs, Sheets, Slides and more. The problem is that they actually already announced this, we knew this was coming, so there isn’t much to delve into especially considering we don’t have access to any of this functionality yet. Here are a few things they did showcase.

Maps is getting an immersive view which showcases the real world, traffic and other details.

Edit photos in the photos app using AI.

Perhaps the most important announcement Google made was the LLM their DeepMind team is working on. If there’s any model that can compete with OpenAI’s GPT-4, I imagine it will be this. The most fascinating thing with Gemini is that they’re training it in such a way that it will be able to reference authors and artists it takes inspiration from when it creates something. This will be huge if it works accurately for protecting creatives’ work. I don’t think we’ll see this anytime soon though so I wouldn’t get my hopes up just yet.

A cure to loneliness

I’ve already written about the impact I think AI will have on love and intimacy. I truly believe that people will become addicted to virtual worlds and as a result, virtual people too. A Snapchat influencer just validated my thought process.

Caryn Marjorie is an influencer. She’s clearly been following the AI wave because she had the brilliant idea of training a chatbot on all her messages and then selling it - Girlfriend as a Service (GaaS) is here now. Why’d she make it? To “cure loneliness” apparently. What a noble crusade.

When I first read the headlines I didn’t think it’d do that well because she was charging it not by the hour or day, but by the minute. She was charging her fans to chat with an AI version of herself per minute. So how did she do? She made $70,000 in her first week. $70k in a week. To put into perspective just how much that is, the median income in the US is approximately $70k.

What we need to remember is this is just a chatbot. This isn’t integrated with VR or AR. There is no visual stimulus. When that happens this will be a lot worse. When VR headsets or even AR headsets can project lifelike people combined with text generation from ChatGPT and audio from Eleven Labs, it’s game over.

Tools

Writesonic

Writesonic is an all in one type platform. You can generate content for a number of different use cases and even upload your own data to train a bot. I haven’t used every feature obviously, but it did work well when I used it.

Bearly

Bearly lets you summarise pages and articles and ask questions on them in an instant. It’s like an AI buddy with you while you browse. I enjoyed using it.

Stable Animation

Stability AI released Stable Animation, a tool which lets you go from text to animation.

If you’re new here you can subscribe to stay up to date

If you’d like to receive more news, you can upgrade to become a premium member

As always, Thanks for reading ❤️ 

Written by a human named Nofil

Reply

or to participate.