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 ‏‏‎ ‎

Let's jump in!

1. Intel gets $8.5 billion to boost domestic chip manufacturing

We start with Intel, who will receive $8.5 billion in grants from the US Government to manufacture semiconductors within the country.

The total amount could increase even further to $11 billion and is part of President Biden’s focus to protect the US from shocks to the supply chain.

Most of the world’s semiconductor chips are manufactured in Taiwan, on the behalf of Intel’s rivals like AMD and Nvidia. But this is a serious risk to the US’ national security, as China wants Taiwan to rejoin the country in the coming years.

If this were to happen, whether by force or through a democratic vote, China would control much of the world’s chip production.

Intel is planning to spend $20 billion alone on their new Ohio facility, with other companies - such as TSMC - lined up to receive billions for building their own manufacturing plants as well.

2. Apple sued by US Government over iPhone monopoly

The US Department of Justice has accused Apple of monopolistic behaviour over its iPhone ecosystem, with the case focusing on 5 specific areas - “super” apps, messaging, cloud gaming, digital wallets, and smartwatch cross-platform compatibility.

It alleges that Apple has suppressed competition in each of the 5 areas, but Apple disputes this and claims that regulators are selectively picking metrics.

This lawsuit has been supported by Apple’s rivals, such as Epic Games and Spotify, who have both brought legal action against the company over similar issues.

Ultimately, we won’t know how this will play out for quite some time. But it will have a major impact on what the iPhone ecosystem looks like going forward.

3. Nvidia’s CEO says AI hallucinations can be solved

The company has been the biggest winner during the AI boom, as its share price has skyrocketed in the last year.

At the Nvidia developer conference, Jensen Huang said that AI hallucinations - which is when they confidently say something that’s completely wrong - could be solved using RAG.

This is a technique called “retrieval-augmented generation”, which involves the model searching for an answer and then using it to respond.

Personally, I seriously doubt this is possible. RAG on its own can’t solve the fundamental issues with today’s generative AI models, as they have no understanding of the world around them.

Even when RAG is used well, they still make mistakes. This is because of how they were developed in the first place.

Although, he is correct in saying that we need other systems and techniques to teach these models about facts, physics, and how the world works.

4. GitHub’s new AI tool can automatically fix code vulnerabilities

For some time now, GitHub has offered code-scanning tools to flag issues - such as mistakenly including an API key.

They were quite basic tools, but the company has announced their new product can now identify and fix security vulnerabilities within your code.

It currently supports JavaScript, TypeScript, Java, and Python - and can detect “more than two-thirds” of security problems.

This will be welcomed by GitHub’s enterprise customers, who face heavy fines for security vulnerabilities and breaches. Although, developers shouldn’t blindly accept the tool’s recommendations - as it may not always understand the code and its purpose.

5. Saudi Arabia plans for a $40 billion push into AI

Developing the latest AI models takes a lot of money, so startups will be excited to hear that Saudi Arabia will be starting a mega fund to invest in the sector.

This is part of the country’s wider goal of expanding their influence and growing in areas beyond the petroleum industry.

While startups will very much welcome the move, governments will be concerned about foreign nations gaining access to their most advanced technology companies.

In recent years, the US has been especially protective of their assets and have blocked takeovers. However, other western countries - such as the UK - are much more lax about this issue.

Microsoft unofficially acquires Inflection AI

The biggest story of the week was Microsoft’s unofficial acquisition of Inflection AI, which was being led by the co-founder of DeepMind.

I say unofficial because Microsoft has signed a deal to hire Inflection’s key staff, including their executives and AI researchers, and also licence their technology - but will not actually merge the two companies.

As we know, regulators do not like to see big tech companies absorb their rivals - especially in critical areas like artificial intelligence research - so this deal seems to be designed to avoid those headaches.

Ultimately, it’s another huge win for Microsoft as they seek to diversify and look beyond OpenAI. It’s especially important given how close OpenAI was to shutting down in November, after their CEO was ousted.

Over the course of the last year, we have seen Microsoft and Google consolidate their positions and partner with AI startups that have emerged as world-leaders.

Microsoft has made deals with OpenAI, Mistral, and now Inflection. Both Google and AWS have made deals with Anthropic, which is seen as OpenAI’s closest rival. While Google has also invested billions in Character.ai as well.

In a short period of time, the disruptors have partnered with the incumbents. This is because the startups need a lot of money and compute to create their AI models, which are two things that the big tech companies have in abundance.

If these startups don’t partner with big tech, they face the prospect of their rivals partnering with them instead - and then using those resources to leapfrog everyone else.

Regulators in the US and EU are already looking into these AI partnerships, but these processes often take a long time and the fines are relatively small compared to the potential upside.

It’ll be interesting to see what they make of this latest deal, now that Microsoft and Google are quickly becoming the most dominant players in AI.

DeepMind develops an AI model to suggest football tactics

DeepMind and Liverpool FC have worked together to develop TacticAI, which is an AI system that provides coaches with insights about corner kicks.

It uses a mixture of predictive and generative AI to analyse the previous plays. Then it suggests what changes should be made to maximise the chances of goal scoring.

Staff at the football club have preferred to use the AI system, rather than practice setups, over 90% of the time.

Trying to predict where the different players will move around the pitch is a tricky task, but the model was able to:

Previously, coaches would have analysed hours of video footage themselves. However, TacticAI takes some of that workload off them.

It’s a really interesting project, which shows what’s possible when you combine both predictive and generative AI for very specific use cases.

🎥 OpenAI tries to court Hollywood with their new AI video generator

🚀 Autonomous robot installs a panel outside the ISS

📦 DoorDash starts to trial drone deliveries in the US

📱 Apple and Google are in talks to bring GenAI to the iPhone

🤖 Stability AI CEO resigns to pursue de-centralized AI

📈 Reddit stock is up nearly 48% on its first day of trading

🔍 NHS uses AI to detect cancer that was missed by doctors

This is a startup based near San Francisco, which is pushing the boundaries of prosthetics with their new bionic arms.

The company is in the early stages of development, but hopes to price the device at around $20,000 - which is considerably cheaper than other options.

Sensors are placed on the person’s limb, which are used by a machine learning model to interpret the electrical signals and then control the prosthetic arm.

There are some remarkable demos that really show how much progress is being made here and the new possibilities it opens for those with disabilities.

In one example, they showed how the arm could be used to pick up items, such as a glass of water and a baseball.

Of course, it’s not the finished product and there’s still some way to go - especially around regulatory filings in the US - but it’s exciting to see technical leaps that can make a huge impact on peoples’ lives.

This Week’s Art

We’ve covered quite a lot this week, including:

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.

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.