AI Regulation Is Coming - Digital Explorers' Diary #40

Welcome to the Digital Explorers' Diary #40!

A curated collection of thought-provoking topics about interactive technologies, AI, web3, sense-making, entrepreneurship, and psychology.

Executive summary:

• Regulations on AI are coming, and they will have implications for your business.
• Meta Quest 3 is promising. Will it have more success in usage than its predecessor?
• Firefly, Adobe’s AI-powered art generation, is now available in beta.
• France will regulate the influencer business.
• And AI-powered «discussions» with dead people.

What’s coming up: Stay tuned for next week’s newsletter, focused on «The Metaverse», with the release of the Apple XR device, the XReal AR glasses, and Nvidia AI-powered avatars. Will next week mark a turning point in the XR industry?

AI regulations and what it means for you.

Over the last couple of weeks, there was a lot of movements around Artificial Intelligence regulation. Let’s dive in.

The European Union AI Act – which is in the work for 2 years, so way before the current hype – has the ambition to regulate AI usage in both private and public sectors. It classifies the AI tools accordingly to their perceived level of risk, with increasing levels of regulation, transparency, and obligations.
In the continuity of the GDPR, the main goal is to protect privacy, for example, by banning any AI that would use biometrics analysis for predictive policy on users.

Then, OpenAI’s CEO Sam Altman went to the US Congress along with IBM’s Christina Montgomery and scientist turned entrepreneur Gary Marcus. The hearing titled «Oversight of A.I.: Rules for Artificial Intelligence» showed a well-prepared Altman: he had answers for everything.
He said that Artificial Intelligence should be regulated, saying, «If it goes wrong, it can go very wrong.»

So what’s going on?

One could argue that Altman’s move is to over-sell the technology by claiming it’s powerful; he drives the hype up and attracts more attention, users, and investors.

But he’s not the only one; many other eminent AI people are warning against the high risks, for example, Geoffrey Hinton, the «godfather of AI,» who recently quit Google for that exact reason.

On the other side of the spectrum, you’ll hear voices claiming that AI is «stupid», so associated x-risks are irrelevant; take, for example, Meta’s VP Yann LeCun, who said, «You cannot invent the seatbelt before inventing the car», alluding that it’s an illusion to try to align AI that does not exist yet.

Anyway, what does that mean for you and your business?

• Regulations will come soon. The EU AI Act has a 2 years grace period, but if you have a business that uses AI, it can mean an increase in compliance costs, similar to what GDPR did.
• Before regulation, education must happen to minimize business risks. Take, for example, this lawyer who thought ChatGPT was accurate and created a filling full of non-existent cases. This can easily be avoided with good training and information. Communicate with your teams!
• Think deeply about your use case. Are you or your clients using AI? What can a limitation in AI usage mean? As an example, look at all the consequences GDPR had, from a rise in overall costs (lawyer, engineer) to businesses looking for new ways to create targeted advertisements. AI is, by the way, helping to solve this. Ha!

In conclusion, here is another thought experiment. Let's assume that AI-powered chatbot search becomes the standard, replacing the standard Google/Bing word-based search. Then, you’ll search for a movie to watch, and the bot will scrape the internet for movies with good ratings, match your preferences, and directly direct you to the streaming service.
Before that, you had to research, watch a few trailers, and read reviews. The websites you were using will then lose their traffic. What will they become? Should they now pivot and find new ways, for example, by creating a ChatGPT plugin? Will OpenAI and Google create new business models for brands and businesses to promote their services in these chatbots?

This must be included in the current regulation discussion, as it will radically change internet business.

This Week's News:

1. Rumors are increasing around the next Meta headset, cleverly named «Quest 3». Set to be released in late 2023, it will bring some of the Quest Pro features to a lower price tag, with some improvements. Much cheaper than the expected Apple device (~$400 vs. $3,000), it obviously targets a different audience. Will it succeed in bringing Meta back as a leader in sales and usage? The Quest 2 has sold ~2 million units, but users don’t use it that much.
   The Quest 2 was a VR headset. With the addition of color cameras, this updated version will be a Mixed Reality device where you can see your surroundings and interact with your environment, especially with the depth sensor that will allow the device to map your room automatically!
   What the Quest3 might look like (Source)
2. It didn’t take long for Adobe to strike back in the Generative AI battle: they released Firefly last week, available in beta for all Creative Cloud subscribers.
   It is embedded into Photoshop, and that’s huge. Generative AI is now embedded into the most popular design tool ever.
   For now, Firefly lets you select an area and prompt it for what you want to see in it or let it fill the area for you.
   As with all AI-powered image generation, it has some successes and failures; it works best on generic «things» than specific ones. You will have more success with a picture of a dog in the mountains than with a double-battery T342 screwdriver. (I made that name up, but you get the gist).
   Adobe Firefly
3. In the «I told you so back in September 22» series, AE Studio is planning to create an AI-powered chatbot that lets you talk with dead people. Brace yourselves for weird times, my friends!
   Powered by OpenAI, you have to feed it emails or chats from that person, and a «persona» is generated and fed into the chatbot.
   As the beta testers mention, the «magic» quickly fades away with the flat tone that is so typical of the current AIs. But, as they get better, the uncanny valley will start to get deeper… Oh, and by the way, if you want to chat with AI, you can try the currently most popular app Replika.
   Related, an influencer launched a virtual, AI-powered version of herself that you can chat with for $1 a minute. And it didn’t go well… obviously.
   I’m usually pretty nuanced in my judgments, but entitled influencers capitalizing on their follower’s loneliness is really not going in the right direction.
   Seance AI
4. Related to this week’s main topic, a law is set to pass this summer in France to regulate the influencer industry. After a series of scams, the law will ban influencers from promoting cosmetic surgery, crypto, financial services, and counterfeit products.
   As usual, the issue will be in the implementation. Who’s responsibility is to apply the law? Should the platforms do it as the E.U.’s Digital Service Act wants them to? If so, will they report and/or ban fraudulent users?

From the Podcast

Every week, I’m teaming up with Guillaume Brincin and Sébastien Spas on the Lost In Immersion podcast.

1. We dive deeper into the topic I discussed last week: will AR glasses replace our computer screens? We agree that it’s the future, but we wonder if the current glasses are good enough for sustainable use. Maybe the yet-to-be-released Xiaomi or the Xreal will change the game?
2. As Meta and Magic Leap are in partnership discussions, our conversation goes into the larger topic of the hype around the potential release of the Apple device and how announcements from the other players in the industry are currently multiplying in anticipation.
3. Seb showed us how NERF technology is getting easier and easier to use. Are you a creator or an agency? You should definitely check the amazing capabilities offered by applications like Luma.ai; it can bring your communications to the next level.

Tune in at 10 am UTC on Wednesdays on Twitch, and listen to the replays on Youtube!