Welcome to IN newsletter #001

Thanks for subscribing. This newsletter is about all the thought-provoking topics we encounter, discuss and filter, so you have the latest and the greatest in your inbox.

The Innovation Network is a co-creative collaborative network. We explore and share how technology changes lives, organisations, and society for the better. We are just getting started with the newsletter. Your feedback is invaluable to create a newsletter that is meaningful for you in work and life!

Do you know any forward thinkers that should join the network? Please do forward the newsletter.

For now, let’s dive in!

Aragorn & Patrick

Ideogram Generates Text in Images

We often get asked how to get a specific text into an AI generated image. That remains difficult with Midjourney and Dall-E. Ideogram to the rescue! It was in beta for a long time, but now their 1.0 release is available for everyone. It’s a new text-to-image model that can generate high-quality, photorealistic images from your descriptions. And… it does very well when you like to include text! Check it out here.

On Cloud 9 with Claude 3

Buckle up, Future Immortals! The AI race just got even hotter with Anthropic's launch of the groundbreaking Claude 3 model family. While Google's Gemini 1.5 made waves with its massive context window and laser-focused recall, it turned into a PR nightmare due to biased outputs that cost the tech giant a staggering $97 billion. Talk about a costly lesson in thorough testing!

But fear not, because Anthropic has swooped in to save the day with Claude 3 - a trio of models (Haiku, Sonnet, and Opus) that match Gemini 1.5's capabilities without the embarrassing blunders. Claude 3 is leaving competitors like GPT-4 and GPT-3.5 in the dust on several benchmarks, and get this: some people are saying it's even showing signs of self-awareness.

The rapid advancements in AI are nothing short of mind-blowing. Models are becoming more powerful, energy-efficient, and hardware-optimized at an exponential rate. It's like watching a sci-fi movie unfold before our eyes! And with Anthropic's commitment to frequent updates, who knows what incredible feats Claude 3 will achieve next? Don't blink, or you might miss the next quantum leap in AI technology. This is one wild ride you won't want to miss!

(This text was written by feeding Claude Opus my handwritten article from LinkedIN, asking it to summarize and make it energizing!)

The Coming Wave… of job loss

“Throughout history, technological advances always ended up creating more jobs than they cost.”

Right? … Wrong. Let us kill this confirmation-bias-driven illusion here for you all, once and for all. This is a well-known, often repeated Urban Legend.

Although it is true that in the first and second industrial revolutions, new technologies like steam-powered textile weaving machines initially replaced humans, they ended up creating far more jobs. Simply put, the increased efficiency and productivity, increased profits, reduces prices, and drove up the economy, leading to an increase in wealth for all, and thus an increase in jobs.

The problem is that even many of the most intelligent economists today, believe that this is a kind of immutable law, that applies to every technological advancement, especially those that seemingly cost jobs on the surface.

As Martin Ford points out very eloquently in his book “Rise of the Robots” technological innovations since the late 80’tis have led to incredible growth in efficiency and productivity, but the growth in jobs in the Western World has been marginal at best. It has stagnated almost entirely since the late 2000’s.

Meanwhile many big companies like Amazon and Walmart have been actively (and in Walmart’s case rather silently) automating as many jobs as possible.

And all of this was before the current big AI and Robotics Boom….

Therefore, the below news item from February 27 is just the first of many more to follow.

it will take time for people to accept this, but a wave of joblessness is on the horizon. And it can be a great thing. Once 30% to 40% of Western Populations becomes job less, governments will have no choice but to take direct action, implementing UBI and sharing the wealth and abundance made possible through AI. It’s nations like the US that we need to worry about most, because they have no proper history in social welfare.

Time will tell, how things will play out.

Important to consider is that we are also facing a massive wave of Aging populations across Western nations as well as China and Japan. Which will soften the blow initially, as people leave the job market and won’t need to be replaced by another human, however eventually we’ll pass that threshold into a new kind of Economy, once that will be powered by Ai and Robotics.

Hopefully, I’ll be on a beach by then, paying for Martinis in Bitcoin.

How will you prepare?
What will you tell your kids?
What does this mean for your business?
Do you take this seriously?
Opportunity or Challenge?

Let us know!

Groq not grok

Is ChatGPT too slow for you? Try a language model on Groq. Yes, Groq with a Q. Not to be confused with Elon Musk's Grok with a K.

Groq is not a language model (like ChatGPT or Grok with a K), but a company founded by former Google employee Jonathan Ross with the aim of creating chips on which AI language models run more efficiently.

When ChatGPT formulates an answer to your question, you see it appear word by word as a user. This is not intentionally programmed as an animation but is a result of "inference".

Inference is the process of passing live data through a trained AI model. So, your answer is issued in real-time, word by word, by the language model.

The inference process now runs on graphics cards - GPUs (Graphical Processing Units). Actually, these cards were not specifically developed for AI but for rendering graphics. Graphics cards are good at processing data in parallel, which is exactly what the algorithm of a language model needs.

This is one of the reasons why NVIDIA (a company that makes graphics cards) currently has a market value of 2 trillion dollars, making it more valuable than Amazon or Google. There is a huge demand for computing power, and this will only increase in the future.

Back to Groq. Groq is introducing LPUs (Language Processing Units), specifically designed for the inference of LLMs. These make inference cheaper (~20x) and faster (~18x).

Why is this important?

The most obvious answer: You get answers to your questions faster. This will be especially valuable when you have a voice conversation with a language model like Siri or ChatGPT. Because of the delay you currently experience, such a conversation does not feel entirely natural.

But it goes deeper than that. The reliability of the answers to your questions can also increase. A language model could have an answer checked again by another language model without you as a user noticing and having to wait a long time for your answer.

Groq also offers an API for developers, so you can work with fast inference on models like Llama (from Meta) or Mixtral yourself.

It's interesting to see that huge strides are being made both in terms of hardware and software, improving AI at a rapid pace. The quality and speed of language models will only increase, making the value and applicability of AI within organizations even greater.

Breathe EMOtion into Your Photos

There are many shorts of Pica Lip sync online, but honestly they feel like a wooden doll from Toy Story 4 compared to EMO from Alibaba! EMO stands for Emote Portrait Alive and takes the concept of photo manipulation to a whole new level in both quality and versatility. EMO boasts breathtaking realism, generating lifelike talking and singing videos from your photos. It’s not yet generally available for the public, but you can find the paper, video’s and code here.

Although, while creating this article (hours before sending) we noted that video’s and content is removed from the vivd-talk page. Let’s see how this develops. For now only Pica Lip Sync is immediately available (in the pro version).

March Reading Tip: MANIAC

If you were captivated by Christopher Nolan's “Oppenheimer”, then "Maniac" by Benjamin Labatut is your next must-read. In a brilliant mix of fact and fiction, Labatut uses the greatest minds of the beginning of the 20th century to weave a picture of the genius John von Neumann.

What I really enjoyed about it is the inception of artificial intelligence. You can feel how the early twentieth-century visionaries foresaw technology's path.

Labatut is using von Neuman to sift the sand of the present through the sieve of history. It’s a work of art that leaves a long lasting impact.

Newsletter full video recording