• The second successful in vivo CRISPR therapy…
  • Synthetic brains?
  • AI is shaking up an entire industry…

Dear Reader,

Last night, about seven million miles from Earth, a spacecraft slammed into an asteroid at 14,000 miles per hour.

It was a real-life practice run for how to deal with a massive Earth-destroying asteroid, as envisioned in the movie Armageddon. It’s not clear if it will work – and it’s odd timing for the experiment considering that’s about what it feels like in the markets right now.

I wrote about NASA’s exciting Double Asteroid Redirection Test (DART) last January when the 1,300-pound spacecraft launched atop a SpaceX Falcon 9 rocket. (Little did we know at the time, but all hell was about to break loose on Earth.)

The target of the DART was a binary asteroid. More specifically, it was a 525-foot wide asteroid (Dimorphos) that orbits a much larger asteroid (Didymos). The goal is to see if the force of the collision is great enough to change the orbit of Dimorphos around Didymos.

Here are the final few seconds before the collision:

Asteroid Collision

Source: New Scientist

DART launched from Earth, orbited around the Sun, went seven million miles, and ultimately hit a 525-foot wide asteroid orbiting around another asteroid.

After all that, we certainly have cause to be impressed. That’s one heck of a long-distance shot.

But it was 55 feet off target. And that will impact the degree to which the collision will affect the orbit of Dimorphos.

It’s not clear yet if it was a success. We’ll likely have to wait a month or two before we can determine if the orbit was changed enough to have “worked.”

And that brings us back to Earth. Fortunately, there is no giant asteroid hurtling toward Earth that we have to contend with right now.

But just like DART, there will be no soft landing if the world’s central banks continue to aggressively raise interest rates, tighten, and implement policies that reduce access to affordable electricity in the middle of a recession.

What we’re experiencing right now is a grand monetary and political experiment that has nowhere near the accuracy of the DART spacecraft.

And as we’re all finding out, the end result is going to be a lot farther off than just 55 feet…

CRISPR genetic editing technology is proving itself in the clinic…

A few days ago, Intellia Therapeutics, one of the “OGs” of CRISPR technology companies, released data from the clinical trial of NTLA-2002.

This is a therapy to treat patients with hereditary angioedema (HAE). And the results were remarkable.

HAE is a genetic disease that causes painful swelling attacks. The culprit is a protein called kallikrein. HAE patients produce too much of this protein. That’s what causes the attacks.

HAE affects about 200,000 people around the world today. That makes it rare. But it’s an awful disease that can sometimes lead to death.

At the moment, there’s only one approved treatment for HAE. The Japanese pharmaceutical company Takeda offers it. Yet it requires patients to go to the hospital for injections every two weeks to manage their swelling attacks.

In contrast, NTLA-2002 is a one-time in vivo therapy. That means the therapy is injected directly into patients. Once inside the body, CRISPR technology edits the patients’ DNA to reduce the production of kallikrein.

Intellia’s premise is that NTLA-2002 can be a one-time cure for HAE. And that’s exactly what the data suggests…

Intellia just treated six patients with NTLA-2002. Half of those patients received a lower dose of NTLA-2002. And the other half received a higher dose.

And the results that just came out demonstrate why these dose cohorts are important.

NTLA-2002 reduced the levels of kallikrein by 65% in the low-dose cohort. And it reduced the dangerous protein by a whopping 92% in the high-dose cohort.

These are both incredible numbers, especially considering this is a one-time treatment.

And that’s why this announcement is so promising.

NTLA-2002 is proving itself to be superior to the only existing treatment for HAE on the market today. And because it’s a one-time treatment, it could be life-changing for patients. Trips to the hospital every other week could be a thing of the past.

And bigger picture, this is further proof that CRISPR genetic editing technology works… not just in a laboratory, but directly in patients.

The success with NTLA-2002 follows on the back of the success that we’ve already seen with Editas’ EDIT-101, the very first in vivo therapy for Leber congenital amaurosis 10 (LCA10), a form of progressive blindness caused by a single genetic mutation.

This bodes well for the future of Intellia Therapeutics. It’s also bullish for the other companies out there working on CRISPR therapies, like those we hold in our Exponential Tech Investor portfolio.

I’m on record saying that CRISPR will help us eradicate all human disease of genetic origin in time. That prediction is proving stronger by the day.

Generative AI just turned its attention to the human brain…

We have been talking quite a bit recently about generative artificial intelligence (AI). These are AIs capable of producing remarkable images based on text input.

In fact, just two weeks ago we talked about how someone won an art competition at the Colorado State Fair with a stunning image he created using the generative AI Midjourney.

Well, this same tech is now being applied to the medical field.

A team of researchers at Cornell University used this technology to produce synthetic images of the human brain.

Here’s a look at the AI at work:

AI-Made Brain Images

Source: Twitter (@warvito)

The AI produces three unique images of the human brain. These are high-quality images that look just like what would result from an MRI (Magnetic Resonance Imaging) of a human brain.

The only difference is, these are synthetic images. They show what a person’s brain could look like.

Of course, that begs the question – what’s the point of this?

And the answer is that this solves one of the biggest problems with deploying AI in the medical field. Today, there are very limited datasets available to train an artificial intelligence. For training an AI, the larger the dataset the better.

The reality, however, is that there is no central repository for medical records and imaging. They aren’t held in one easy-to-access centralized database. This is why if we switch doctors or hospitals, they have to call our previous doctor to get our medical records.

What’s more, hospitals can’t share or aggregate all the imaging data they have. That’s due to privacy laws.

So there’s a mountain of data out there that the industry simply can’t use to train an AI.

The result is that existing datasets tend to be very small. And that limits the capabilities of AI within the medical field.

The researchers at Cornell are solving this problem for brain imaging. Their AI can easily produce 100,000 unique brain images in a matter of seconds.

Again, these aren’t real images. But they could be. They are representative of the human brain.

In fact, the AI can even produce brain images according to certain parameters. This includes images of human brains at different ages or across genders.

And the researchers will package all these images into a massive dataset to train AIs.

This will enable the medical industry to use AI to identify problems and potential solutions when analyzing real MRIs of the brain… and perhaps even gain greater insight into the human brain, how it works, and how it develops over time.

Generating a large, synthetic dataset for training an AI isn’t new. Other fields have done this. But this is the first time medical imaging will use this. And this will immediately solve one of the biggest bottlenecks to making an AI a useful and accurate tool to augment human experts.

Years down the road, we’re going to look back on 2021/2022 as a time when AI hit an inflection point in medical science and biotechnology… and led to a remarkable number of scientific breakthroughs and therapeutic development.

Generative AI is already upending a major industry…

Staying on the subject of generative AI – something else remarkable just happened.

In the span of just a few weeks, this technology has completely upended the stock photography industry. It’s remarkable how fast this is moving.

For context, the stock photography industry provides high-quality images for companies and institutions to use for their own purposes.

Companies often use these photos for marketing applications. But people can use them for branding and creative purposes as well.

For example, if we watch any marketing or promotional video, we’ll likely see numerous stock photography images without realizing it. We’ll find these photos plastered all over modern websites as well.

Up until this point, humans produced all these stock photo images. But that just changed. Check this out:

AI Stock Photos560

Source: Adobe

I just went to Adobe’s stock photography site and searched for “AI-generated” images. There are 61,405 images now in this category.

What’s amazing is that it’s only been a matter of weeks since generative AI like Midjourney and OpenAI’s DALL-E 2 have been available for general use.

All of these images likely appeared on the site in just the last several weeks. They didn’t exist previously.

And look at how remarkable they are.

Adobe is obviously pushing Halloween-themed images to the top of its search results right now. And these are quite stunning. I’m sure if we took the time to scroll through the other 61,000-plus images, they would be fantastic as well.

And it’s not just Adobe.

Popular stock photography company Shutterstock now lists 20,811 AI-generated images on its platform. And iStock has an incredible 88,587 AI-generated images available.

So we are seeing an entire industry transformed in a matter of weeks. That’s all because these text-to-image AI tools are now available for public use.

I don’t think I’ve ever seen new technology completely upend an industry this quickly.

Of course, traditional artists and photographers aren’t too happy about this. Some are surely in panic mode. How can they compete with these stunning images that the AI can produce in mere seconds?

Simply put, those individuals who have made their careers in this industry will need to adopt this technology or adapt to the new climate.

A few days back, we talked about a new job category – prompt engineers. This is a unique skill set of providing a generative AI with the right set of words to elicit a unique generative output.

The reality is that AI results in human augmentation. It empowers us to do more with less effort, and we can control the inputs and modify the outputs to whatever it is that we’re trying to attain.

Some will resist this transition.

But as the industry has already gravitated toward open-sourcing these core AI algorithms, I’m afraid the black cat of Halloween is already out of the bag…


Jeff Brown
Editor, The Bleeding Edge