Digital Twins | Is AI all of us? | Let’s Meet | OpenCRISPR-1 | Astribot and Poe

Hello, Lovely innovators, Trailblazers, and Forward thinkers,

What an exhilarating time to be alive!

Following our latest bi-weekly editorial meeting (where we dive deep into emerging trends) it's impossible not to feel both awe-inspired and slightly bewildered by the rapid pace of change affecting us and future generations.

This underscores the importance of engaging in meaningful dialogues about these trends and sharing our insights with the world.

So… Don’t miss out! Join us for our next in-person gathering on Monday, May 13th, at the High Tech Campus in Eindhoven. Are you passionate about trends, society, AI, innovation, and meeting wonderful people? Then secure your spot now.

In this edition of our newsletter, we delve into the fascinating world of digital twins. (no, Paul, this text is NOT AI generated)

I’m still captivated by Reid Hoffman’s video of his Q&A with his own digital twin. How long until you have your own twin managing your online meetings? Are Microsoft's Phi-3 Mini or Apple’s OpenELM up to the task, or will they ‘only’ revolutionize how we search for data on our devices? Mustafa Suleyman has an interesting take on it. Read on to learn more.

For those who love discussing these cutting-edge developments, mark your calendars for June 12th. We’re hosting an exclusive innovation dinner at Proeflokaal Katendrecht in Rotterdam, where we'll explore technological insights and strategies for navigating them. It’s an opportunity to connect and learn together in an intimate setting. Be there, or invite a friend who could benefit! More info soon!!

Lastly, we value your feedback! Let us know what you think of the newsletter, share your ideas, or express interest in contributing. Also, be sure to follow us on our newly launched LinkedIn page to join ongoing discussions.

Without further ado, enjoy this edition of the Innovation Letter!

Warm regards,

Aragorn & Patrick

The Astonishing Reality of Digital Twins

Ok, check below video from Reid Hoffman and… well… Reid Hoffman’s twin, better known as Reid AI. I think many of you envisioned agents like this somewhere down the line, but to see an actual recording of Reid having a conversation with himself is astonishing. 

Reid AI is generated by Hour One; its voice was created by Eleven Labs. His knowledge is generated from a custom chatbot built on GPT-4 that was trained on Reid's books, speeches, podcasts, and other content that he has produced over the last few decades.

Why did Reid deepfake himself? Partly, I want to discover if conversing with an AI-generated version of himself can lead to self-reflection, new insights into my thought patterns, and deep truths. Don’t we all have some mental feedback if we see a video of ourselves?

What if you can converse with yourself? What would you ask you?

Reid’s conversation with himself is only the tip of the mountain regarding AI-powered Digital Twins and how they will change our lives. This is why Aragorn wrote an article about our Future with Digital Twins, the benefits, and some of the risks involved.

Is AI all of us?

Have you ever pondered what AI truly is? Do you have a definition or perhaps a metaphor for it? Mustafa Suleyman, currently the CEO of AI for Microsoft, writer of ‘The Coming Wave,’ and co-founder of DeepMind, offers his insights in a compelling TED Talk.

Suleyman proposes that we are on the brink of a new era in artificial intelligence. Rather than merely viewing AI as a tool, he suggests thinking of it as a new form of life—a digital species—that is rapidly evolving and gaining autonomy. He introduces a fascinating perspective: could AI represent a collective consciousness?

What resonates with me is his idea that a Large Language Model (LLM) isn't just an aggregation of knowledge; it's animated, intelligent, and interactive. Does an LLM represent the collective 'mind' of a group, a company, or even a nation?

JOIN US THE 13TH OF MAY FOR THE INNOVATION NETWORK MEETUP!

SIGN UP NOW for our first meet-up!

Innovation Network aims to bring together experts, executives, and everyone else who is focused on innovation.

This get-together is open to all members of the Innovation Network. We like to bring you all together to get to know each other. The agenda for the day is straightforward:

16:00 Opening & Welcome
16:30 Panel discussions with all of us in round tables
18:30 Closing drinks

This will be a fantastic opportunity to get to know each other and discuss various topics, from the far future of technology to projects that keep us awake today.

Where? High Tech Campus in Eindhoven. Hosted by our great friends from 3EALITY on the High Tech Campus in Eindhoven.

AI Revolutionizes Human Genome Editing

In a groundbreaking development, researchers at Profluent Bio have successfully edited the human genome using an AI-generated gene editing system, OpenCRISPR-1.

This achievement surpasses the capabilities of the widely used SpCas9 gene editor and highlights the rapid advancements in AI-driven biological design. The Human Genome Project, a landmark international scientific research project, took 13 years and cost approximately €3 billion to sequence the human genome.

In stark contrast, Profluent Bio's AI-powered platform can generate numerous gene editing systems at will, with OpenCRISPR-1 being just the tip of the iceberg.

"The implications of this technology are truly mind-boggling," said Dr. Stephen Nayfach, another research team member. "We can now design and test gene editors in hours, something that would have taken lifetimes to explore experimentally."

This remarkable feat showcases AI's potential to revolutionize gene editing and pave the way for new therapeutic possibilities. As we grapple with the ethical implications of this technology, one thing is certain: the future of human genome editing is here, and it's powered by AI.

Astribot Will Blow Your Mind

Move over, Boston Dynamics; there's a new speed demon in town! The Astribot S1 humanoid robot redefines what's possible with its lightning-fast reflexes and smooth, lifelike movements.

This bot can execute actions at a blistering 10 meters per second! To put that in perspective, the average human can only run at about 15 mph. So the Astribot S1 could literally run circles around you.

But don't worry, it's not just fast, it's also strong. Each of the S1's arms can handle a payload of 10 kg, so it could probably bench press you with one arm while running a 100m dash with the other.

The video of this bot in action is simply mesmerizing. It can pluck a tablecloth out from under a stack of wine glasses without spilling a drop. Can you do that? 🙂

So, if you ever need a robot to run errands at warp speed, grab heavy stuff, or clear a table like a pro, the Astribot S1 is the one for the job. Don't challenge it to a footrace unless you have some serious wheels. This humanoid is no joke!

Size. It’s all about Size…

Recently, I had the pleasure of sharing the stage with Peter Hinssen, a renowned Belgian entrepreneur, writer, and thinker. A few years ago, he wrote a book called ‘The New Normal’. Today, he talks about the ‘Never Normal’. What does he mean by this?

In past newsletters, I talked about the Gartner Hype Cycle and how it is broken for AI. We are now moving from one breakthrough to the next, with barely weeks in between.

As a result, there are no moments anymore between technological breakthroughs and innovations to take a breather.

Again, in the past two weeks, this has been shown by the release of not one but multiple incredible AI models. This time, what was most impressive was not the capabilities of these models but their size.

Meta released Llama-3, and only days later, Microsoft released Phi-3-mini.
Both models operate at close to GPT-4 levels.

So why is this such a big deal? Size. It’s all about Size.

GPT-3.5 has tens of billions of parameters.
GPT-4 and Claude 3 have approximately 2 trillion parameters.

It goes without saying that this shows the exponential growth of these models. However, their size also made it impossible to run these models locally on a PC, Laptop, smartphone, or IOT device.

Their size and subsequent need for computational power also meant significant power requirements.

Llama-3 and Phi-3-Mini, on the other hand, only have 3.8 billion parameters, yet still manage to approach GPT-4 performance levels.

This means these models not only reduce compute and energy requirements by orders of magnitude. They can also be run locally on a Smartphone, changing the game entirely.

Running these models locally means no dependency on the Internet and much faster response times. Bringing us one step closer to permanent, real-time AI augmentation of everything we do.

Check out the VERGE article on it if you want to know more:

AI Tip: POE

You’ve heard us ramble on about Perplexity.ai, but you may not have heard of another interesting player on the market yet. Poe

Poe, I suspect named after that Artificial Intelligence from Altered Carbon (book series + Netflix Sci-Fi show), is a platform that like Perplexity, provide you with access to multiple Ai models.

Poe’s paid version gives you access to almost every model out there. The list is very long.

Personally, we find the choice too overwhelming. However, Poe is your best bet for checking out all the different models without having to go to 100 different websites and pay multiple subscriptions.

In addition, they offer an interesting feature that allows you to chat with multiple AI models simultaneously.

At this point we are still experimenting with this, but expect us to tell you more about this in weeks to come! of go check it out yourself ;)

Newsletter full video recording