SpaceX makes history with re-usable rockets

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 ‏‏‎ ‎

• Why the FBI secretly created a cryptocurrency

• Anduril’s new attack drone that autonomously monitors targets

• Tesla’s plans for autonomous taxis and robots

• … and much more

Let's jump in!

1. AI pioneers win Nobel Prizes in Chemistry and Physics

DeepMind’s CEO Demis Hassabis and Director John Jumper have been awarded the prize for their work on "protein structure prediction".

AlphaFold is an AI model that can predict the 3D structure of proteins, using only their genetic sequence.

Previously, this process took years of lab experiments. But thanks to DeepMind’s progress, it has been shortened to just a few hours.

This breakthrough has significant ramifications for drug discovery, disease diagnosis, and bioengineering - with DeepMind continuing to make progress as they update the AlphaFold model.

The announcement followed the Nobel Prize in Physics, which was also awarded to AI pioneers.

Geoff Hinton and John Hopfield won the award for their significant research into machine learning.

2. FBI secretly created a cryptocurrency to investigate fraud

The FBI has revealed that it created NexFundAI, an Ethereum-based cryptocurrency, as it investigated price manipulation in the crypto markets.

The investigation has led to the SEC charging 12 people, who are suspected of using schemes to artificially boost the prices of crypto assets.

The Department of Justice (DOJ) has also taken action and charged 18 people for "widespread fraud and manipulation”.

Three market makers - ZMQuant, CLS Global, and MyTrade - allegedly wash traded on behalf of NexFundAI, without realising that it was created by the FBI.

As a result, $25 million has been secured from "fraudulent proceeds". The DOJ will return this money back to investors.

3. Big Tech is betting on AI agents

Tech companies are investing heavily into AI agents and researching how they can be used by businesses.

Essentially, these are teams of AI bots that can perform tasks, make decisions, and communicate with each other - without the need for a human to constantly watch them.

Microsoft, Google, and OpenAI are all developing their own versions of AI agents.

However, most have only released tools that include one AI agent - not multiple agents.

I have developed several AI agent tools in the last year, which are incredibly easy to get working.

The bots are able to figure out how to complete a task, without you explicitly stating how to do so.

They’re very impressive, but still somewhat limited in what they can do.

We need further performance improvements from LLMs, before we can realise their true potential.

4. Amazon uses computer vision to improve deliveries and spot vehicle defects

The company is using computer vision to determine when a package should be delivered.

As the driver stops at a location, a green light is shown below the package - eliminating the need for drivers to search through their vans every time.

Early results suggest that the system has significantly reduced the time spent at each stop, which has fallen from 2-5 minutes to under 1 minute.

Following several years of tests, the technology will be deployed on 1,000 electric vans in 2025.

The company also scans each vehicle as it enters and leaves the depot.

Dozens of photos are taken of the vehicle to spot any damage, including scrapes along the side or broken tail lights.

When these are detected, the vehicle is grounded and automatically scheduled for a repair.

This also saves Amazon a lot of time and money, since staff don’t have to manually check the vehicle at the start and end of their shift.

5. Anduril unveils their new attack drone

The defence startup has revealed its latest product, the Bolt-M. The small drone can work autonomously to monitor military targets and attack them.

Unlike the drones that are operated by humans in Ukraine, the Bolt-M uses machine learning to track vehicles and infantry.

It can be used for over 40 minutes and has a range of 12 miles, with the ability to carry warheads that weigh up to 3 pounds.

With the war in Ukraine, attitudes towards AI-controlled weapons are shifting.

Previously, agreements were tabled to ban the weapons entirely. However, this has faded away as the West responds to the invasion and Europe begins re-arming.

Tesla reveals its plans for autonomous robotaxis

The company has revealed the Cybercab, which is a two-seater autonomous vehicle and doesn’t include a steering wheel or pedals.

Elon Musk has promised that the Cybercab will be in production “before 2027”.

The Robovan was also unveiled and this autonomous vehicle can hold up to 20 people - or it could even be used for deliveries.

Interestingly, Tesla’s humanoid robot also made an appearance. At the event, it was able to pour drinks and pose for selfies.

On the face of it, you might think those announcements are impressive. And they should be.

However, none of the technology was fully autonomous - which was very deceptive. Instead, humans were behind the scenes and controlling Tesla’s systems.

The humanoid robots were not autonomous. They were not replying to the attendees, it was simply a human controlling it from behind the scenes.

In a similar vein, the Cybercab had a very controlled demo. It only drove around several small streets at the Warner Bros studio.

Personally, I see these announcements as a vision for the distant future. But nothing more.

It’s very disappointing. Following the event, Tesla’s stock price has fallen by over 10%.

That’s partly due to the difficult market for tech stocks, but it also indicates how little confidence investors have in Elon Musk’s promises.

SpaceX makes history with re-usable rockets

We shift from one Elon Musk company to another, but this one is actually grounded in reality (no pun intended).

SpaceX has made history, as their Super Heavy rocket returned to earth and was “caught” by huge mechanical arms.

This is a huge milestone and brings us a step closer to re-usable rockets, which could significantly reduce the cost of space travel.

Yesterday, the spacecraft launched from SpaceX’s facility in Texas.

Minutes later, the two rocket components separated and the booster rocket quickly returned to earth.

The booster was brought down in a controlled way, which slowed down enough for the mechanical arms to grab the rocket.

It took just 7 minutes for the entire process.

📱 TikTok lays off hundreds of employees, uses AI for content moderation instead

🚀 SpaceX alumni are raising $550M for a deep tech fund

🔓 Fidelity says 77,000 people are impacted by data breach

⚖️ US government could break-up Google, but a fine is more likely

✈️ Boeing cuts 17,000 jobs, which is 10% of its workforce

🍎 Apple's study proves that LLMs cannot reason

Qantev

Qantev is a tech startup that’s focused on health insurance. It has raised over $45 million dollars in the last few years and uses AI to automatically process claims.

They’re already working with huge insurers, including AXA, Generali, and Bupa.

What’s interesting about their strategy is that they don’t use Large Language Models (LLMs).

Instead, they prefer to use smaller AI models and then customise these models for the health insurance sector.

This helps to boost accuracy when dealing with claims and marks them out from the competition.

The company is based in Paris, but has plans to expand into Asia and North America. There’s also the potential to move into other areas, such as underwriting.

If you want to find out more about the startup, I’ve added a link below.

This Week’s Art

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

• The AI pioneers that have won Nobel Prizes in Chemistry and Physics

• Why the FBI secretly created a cryptocurrency

• Big Tech’s bet on AI agents

• How Amazon uses computer vision to improve deliveries and spot vehicle defects

• Anduril’s new attack drone that uses machine learning to track targets

• Tesla’s plans for autonomous taxis and robots

• SpaceX’s historic achievement with re-usable rockets

• And why Qantev have decided against using LLMs

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 AI Engineer and works within Kainos' Innovation team. He identifies business value in emerging technologies, implements them, and then shares these insights with others.