🔬 AI tool predicts Alzheimer's 7 years before symptoms appear

Welcome to this edition of Loop!

To kick off your week, we’ve rounded-up the most important technology and AI updates that you should know about.

‏‏‎ ‎ HIGHLIGHTS ‏‏‎ ‎

• The startup that’s adding AI to fighter jets

• Why TikTok faces a nationwide ban in the US

• DeepMind's new AI that can play video games

• … and much more

Let's jump in!

1. AI tool can predict Alzheimer's before symptoms appear

We start with some positive news from UC San Francisco, where researchers have created a new tool that can predict if someone is likely to develop Alzheimer's.

They achieved this by analysing over 5 million patient records for different risk factors. The tool performs surprisingly well, as it was able to accurately predict it for 72% of people.

By catching the disease early, doctors can prescribe medications that could slow down its progress and lessen the impact.

The researchers hope to extend their methodology to other challenging diseases, like lupus and endometriosis, as well.

2. The House passes a bill that could ban TikTok

The US House of Representatives has swiftly passed a bill that would force TikTok's Chinese parent company to sell the app, or face a nationwide ban.

But the bill faces a less certain future in the US Senate, where it will likely be slower to progress.

Both political parties are united against TikTok being owned by a Chinese company, as they fear the extensive data collection could be used in the future to target Americans.

Over 100 million people in the US use TikTok, which provides unparalleled insights into their general interests, personality, and political opinions.

It’s tricky to predict how this will play out. China passed a law in 2020 that prevents sensitive tech from being sold to foreign companies, which would mean an American buyer of TikTok wouldn’t actually get their hands on the app’s algorithm.

That makes the sale a less attractive prospect, as it’s seen as the app’s “secret sauce” and why users are hooked on it.

It also raises the possibility that potential buyers could get caught between the US and China. That’s particularly difficult for many big companies, as they often have their own operations in China and don’t want to put them at risk.

Right now, everything points to TikTok being sold to an American buyer - but they’re unlikely to get everything they want.

3. Apple buys an AI startup that specialises in small models

The company has acquired a Canadian startup called DarwinAI, which uses computer vision to monitor components during the manufacturing process.

Their software is used by manufacturers to improve efficiency and spot issues with the product.

It’s possible that Apple could use this technology to improve their own device manufacturing, but the company already has a good reputation in this area.

What’s interesting here is how DarwinAI has specialised knowledge on how to make AI models much smaller and faster.

This is critical for Apple, as they aim to bring more generative AI features to their devices and catch up with their competitors.

Just a few weeks ago, their CEO Tim Cook spoke about how they will “break new ground” this year with generative AI.

4. Waymo will launch their robotaxi service in Austin later this year

While Cruise’s self-driving operations have imploded in the last few months, Waymo is slowly expanding theirs.

They’ve announced plans to launch their robotaxi service to Austin, Texas by the end of the year - which follows a similar launch in Los Angeles this week.

I was at a talk by Waymo’s Co-CEO, Tekedra Mawakana, and was impressed by how they’re approaching the challenge.

Waymo is much more realistic about the need to slowly introduce the service, which contrasts with their competitors. They’re working closely with regulators and local residents to build trust in the technology.

The company is also interested in expanding into Europe, but there are no plans to announce at this stage.

The UK has recently announced their own framework for self-driving vehicles, which should make it a more attractive area to invest in.

5. EU questions tech giants on AI risks to elections

The European Commission has sent formal requests for information to Google, Meta, Microsoft, Snap, TikTok, and X.

Under the new Digital Services Act, these platforms are required to assess and address systemic risks - such as those related to AI-generated misinformation, deepfakes, and voter manipulation.

The Commission is particularly concerned about the potential impact for the upcoming European Parliament elections. It plans on conducting stress tests that will evaluate the platforms' readiness for these challenges.

The EU is also focusing on broader risks, such as explicit deepfakes and malicious synthetic content.

DeepMind’s latest AI model can play 3D video games

A few weeks ago, I outlined how DeepMind developed a GenAI model that can create 2D video games from text.

Well, they’ve just unveiled progress on a different model that can actually play 3D games.

SIMA is capable of playing games like a human and can follow verbal instructions.

The model is able to associate visual representations of actions, objects, and interactions - meaning that it can generalise across different games.

There have been hints at this for some time. Microsoft unveiled MM-Vid late last year, which could play simple games like Super Mario Bros.

For Google, this provides them with a new product opportunity. As I’ve raised before, we will likely see companies offer AI tools that can play video games with you.

At first, that sounds a bit weird - normally you want to play and experience it with your friends or family members. But there are plenty of times when people are gaming by themselves.

If you can have an AI companion that substantially improves your overall gaming experience, then it might be something that people will be willing to pay extra for.

Head of ChatGPT says companies will shrink due to AI

I was at SXSW last week and watched Peter Deng, who oversees ChatGPT at OpenAI, speak about where the technology is heading.

Given it was SXSW - which involves creatives spanning art, television, movies, and music - it was predictable that there would be questions about compensating artists for their work, which are currently being used to create their huge generative AI models.

Surprisingly, Peter Deng wasn’t prepared and avoided answering the question directly - instead asking the audience for their view. As you’d expect, they supported the artists.

There are reasons why Deng couldn’t comment on this, specifically the lawsuit that has been filed against OpenAI over this very issue.

To date, OpenAI and other big tech companies have argued that fair use allows them to do this - without any compensation or credit.

Deng did admit that companies will get smaller, but that “there will be more of them”. I’m not sure I agree with this.

Strangely, most people have missed the head of ChatGPT’s inadvertent admission that job losses are coming.

Alarm bells should be ringing for politicians and regulators, as they face the prospect of growing unemployment levels and the wide-ranging consequences that will follow for their economy.

Over the last year, we have looked at OpenAI as a world-leader in Artificial Intelligence - which they certainly are.

But this moment was a reminder that they’re a young start-up, without clear views on how their technology will impact wider society.

That’s fine, as it’s the job of regulators to outline how the technology should and - more importantly - shouldn’t be used. It’s time that nations got serious and actually plan for a future that isn’t far away.

🎥 OpenAI will release Sora, their AI video generator, later this year

🦾 Mercedes are piloting humanoid robots from Apptronik

🚀 SpaceX makes significant progress with their Starship test flight

📝 Bear Robotics, who are developing robot waiters, raise $60M from LG

🎮 LinkedIn plans to add gaming to its platform

Helsing

I’ve touched on Helsing very briefly before, but haven’t covered them in much depth.

They’re an impressive defence startup that’s based in both London and Germany, which has raised over $223 million of funding in their last round (series B).

Helsing are one of the top AI defence startups in Europe and have signed a contract to upgrade Germany’s Eurofighter jets with AI capabilities.

Specifically, their AI platform will analyse the plane’s radar data and use this to generate self-protection measures that will prevent it from being spotted by enemy radar.

Last month, they signed a deal with Ukraine to provide them with access to their latest technology.

We’re starting to see huge amounts of money invested in defence startups, such as Helsing and Shield AI in the US, who are making the case that we need to protect our democracies by developing new AI capabilities.

With Russia’s invasion of Ukraine and tensions rising with China, they might not be wrong. You can read more about Helsing using the link below.

This Week’s Art

We’ve covered a lot this week, including:

• How an AI tool can predict Alzheimer's 7 years before symptoms appear

• US House of Representatives passing a bill that could ban TikTok

• Apple’s acquisition of an AI startup that specialises in small models

• Waymo’s plan to launch their robotaxis in Austin

• EU concerns about the risks AI poses for elections

• DeepMind’s AI model that can play 3D video games

• Head of ChatGPT’s thoughts on the future

• And how Helsing are adding AI capabilities to fighter jets

Have a good week!

Liam

Share with Others

If you found something interesting in this week’s edition, feel free to share this newsletter with your colleagues.

About the Author

Liam McCormick is a Senior Software Engineer and works within Kainos' Innovation team. He identifies business value in emerging technologies, implements them, and then shares these insights with others.