• Your new doctor is… an AI?
  • Google wants to end online passwords
  • This AI can read our minds

Dear Reader,

Waiting for the next generation of electric vehicle (EV) battery technology has been a frustrating endeavor for investors who are tempted by the thought of a battery breakthrough.

Unlike most areas of high tech over the course of the last three decades, battery technology just hasn’t improved at an exponential rate. The same underlying technology used in lithium-ion batteries in 1995 is still used in today’s consumer electronics and electric vehicles.

The energy density, quality, and charging rates have improved incrementally every year; but surprisingly, there have been no major breakthroughs.

The last five years brought hope, however, with a large number of private and public companies pursuing different forms of solid-state batteries as the potential solution to this multi-decade conundrum. Many made aggressive claims that they cracked the engineering problem, and all that was left was to scale for manufacturing.

A perfect example is a company that I’m sure many have heard of – QuantumScape (QS).

It claimed that its revolutionary lithium-metal technology was the answer to the next generation of battery technology that would replace traditional Li-ion batteries. Among other things, QuantumScape promised:

  • Lower manufacturing costs compared to Li-ion batteries

  • 82% improvement on energy density

  • An EV range of 450 miles

  • 15-minute charging time from 0-80% full

It all sounded fantastic. And QuantumScape was targeting a potential market worth $450 billion in annual sales. What’s not to like?

There was enough hype and excitement back in 2020 around QuantumScape’s message that it went public at a valuation of $3.3 billion via a reverse merger.

I remember first analyzing the company and scratching my head. It wasn’t forecasting any meaningful revenue until 2026, and it was touting a “reasonable” valuation of “just” 1 times 2027 annual sales.

Even better, the CEO of QuantumScape pitched a future enterprise value of $45 billion based on 2028 estimated revenues (reminder: this was back in September 2020).

I didn’t buy it. There were red flags everywhere. But incredibly, some did.

QuantumScape came out of the gates at $10 a share as most SPACs do and rocketed to $131 a share for a very brief moment. What happened since then has been a slow-motion train wreck. The stock now trades at just $6 a share, 40% below the IPO price and more than a 95% drop from its highs.

And the last few weeks have been nothing short of a capitulation by the company.

Announced earlier this month, QuantumScape is pivoting from its aspirational next-generation EV batteries to focus on smartphone-sized batteries. Apparently, scaling up to manufacturing turned out to be a lot harder than the company let on.

But it is actually worse than that. QuantumScape has struggled with both its lithium-metal anode and its ceramic separator, which were two of the things that the company was supposed to have its head around.

QuantumScape may look “cheap” right now at $6, but it still sports a $1.78 billion valuation which makes zero sense at all. Even presuming things go well with the pivot, I doubt the company will have any significant revenue until 2026.

And while it is sitting on $983 million in cash, it will almost certainly burn through all of it by 2025, resulting in the need to raise additional capital or sell itself for a fraction of the price.

As it turns out, the self-proclaimed leader is far behind. Other players like SES AI, Enovix, Sila Nanotechnologies, and Group14 Technologies are showing much better near-term promise.

But next-generation battery technology is a tough problem to solve for every company. The research and development costs alone are in the billions. Even SES AI and Enovix will have to raise more capital despite the progress that they’ve made.

And solid-state battery technology, at the scale required to “fuel” electric vehicles, has still yet to be proven.

Generative AI versus human doctors…

A team of researchers from the University of California San Diego and Johns Hopkins University just published a study that provides an interesting lens on the future use of AI in healthcare.

The research took an early look at the employment of generative artificial intelligence (AI) chatbots. Researchers used an online forum and allowed chatbots to answer medical-related questions from patients. Then they gathered 195 exchanges. The results may surprise us…

The study analyzed the difference between the responses from human physicians compared to exchanges with medically trained AIs. I love it. There’s a lot we can learn from studies like this about the future of AI in healthcare.

Check this out:

It turns out that the humans scoring these responses preferred the interactions from the AI rather than the human doctors 78.6% of the time. And as we can see in the graph above, the quality of responses from the AI turned out to be about 3.6 times better than the responses they got from the physicians.

What’s more, the human respondents felt that the AIs were 9.8 times more empathetic. The panelists reviewing the responses were impressed that each response from the chatbot was personalized and went beyond “boilerplate answers.”

In one example, an online patient asked about what happens if they swallowed a toothpick. The response from the human doctor was 58 words in total, while the chatbot answered with 191 words.

To me, this is very encouraging research. And it’s also interesting that humans felt they had a better overall experience speaking with an AI rather than a physician. It’s even more ironic that the AI (software) demonstrated far better empathy than humans did.

This is huge. And what’s exciting to me about the use of a medically trained AI is that the whole world will benefit. The AIs can work 24/7, 365 days a year. They don’t need to eat, drink, or sleep. They’re always available, and they never get tired or complain.

But even more than that, we could train a single AI on the entire body of medical knowledge available to us. That includes all the different specialties in medicine. It goes without saying that this would be an impossible task for any human doctor.

Using AI in this way will go a long way toward alleviating the shortage of physicians around the world, especially in developing countries. That shortage is further amplified when it comes to specialists.

Imagine if we had a single AI that could diagnose, prescribe, and interact with patients about any medical matter. And the AI could do so at a very low cost.

This technology could completely democratize access to world-class healthcare. Anyone on Earth with access to a computer or smartphone could get an appointment with an AI instantly. Even diagnostic data could be provided to the AI such as results from blood tests or imaging.

This dynamic would reduce the overall burden on human physicians. And it would provide a better experience for patients as we’ve learned above.

This technology isn’t a replacement for medical professionals. It is a force multiplier that will ultimately reduce the burden on healthcare systems around the world and result in dramatically better outcomes for patients.

Passwords are about to become a thing of the past…

Google just made an announcement that’s set to completely streamline the way we access all of our various accounts online.

Most of us are used to having scores of different login names and passwords for all kinds of online accounts. We have our online bank accounts… our utilities accounts… our credit card accounts… our insurance accounts… our mortgage accounts… our social media accounts… and various subscription-based services that we access online.

For years now, we’ve had to manage all these different accounts separately. It’s remarkable that the same system has been in place for 20 years.

What little innovation has happened, like the use of two-factor authorization, was a little over a decade ago. That tech sends us text messages or emails with secret codes every time we log into one of our accounts.

The idea was to verify our identity and add an extra level of security. But the problem is, those forms of authentication aren’t terribly secure. Anyone with access to our phone or email could intercept our authentication codes.

The next evolution of this came in the form of several different apps that constantly generate a new log-in code every 30 seconds for each internet account that we have. This is a better form of authentication in that it is more secure… but it still requires us to jump through a lot of hoops just to log into a single account.

Google’s big announcement is going to change all that.

The tech giant just launched a system that will make passwords and authentication codes a thing of the past. The new system produces a secure cryptographic code on our smartphones. And it combines this code with some form of biometric identification. That may be facial ID. A fingerprint. And in the near future, it may be voice authentication.

So this new system provides us with incredibly secure multi-factor authentication – no passwords or codes necessary. It will make signing into all of our online accounts quick and simple.

Here’s a look at how the system works:

Here we can see how simple Google’s new system is. We just enter our email address and complete our biometric ID – a fingerprint in this case – and we’re done.

Several major companies have already adopted Google’s new system. This includes DocuSign, PayPal, Shopify, and Yahoo. And I expect we’ll see many others follow suit in the months to come.

This is the future of account security.

We’ll likely see other companies launch their own version of this authentication system as well. Google won’t be the only game in town.

I certainly look forward to this technology being implemented. Fortunately, the days of managing hundreds of logins and passwords could soon be over… and without sacrificing security.

An AI that can turn our thoughts into text…

We’ll wrap up today with an incredible breakthrough in the generative AI space. New research out of the University of Texas at Austin shows that generative AI can read human brainwaves with precision.

Back in March, we explored some similar research coming out of Japan. A team there used functional MRI (fMRI) imaging technology to capture human brainwaves while people were looking at pictures.

They then fed those brainwaves to a neural network… and it was able to recreate the images that people were looking at. It was pretty incredible to accurately recreate images simply from the brainwaves recorded in an fMRI scan.

This research takes the technique one step further…

The Texas research team used fMRI machines in their research also. But instead of showing people images, they showed them a silent movie and recorded their brainwaves.

The team then enabled generative AI with what’s called a semantic decoder. This enabled the generative AI to translate human brainwaves into a text description of what the patient was viewing. And it was accurate enough to explain what the viewer was seeing…

For example, here’s a snapshot of the text produced by the AI: “I got up from the air mattress and pressed my face against the glass of the bedroom window expecting to see eyes staring back at me but instead finding only darkness.”

The AI translated that text directly from one of the participant’s brainwaves. And it describes perfectly what the person was thinking as they watched the silent movie.

So generative AI can be used to read our minds. That’s wild. And there are some incredible potential applications…

There are plenty of people out there who have lost their speech due to a stroke, trauma, or spinal cord injury. This technology would enable them to have discussions with their family members again.

And suppose law enforcement had a key suspect to a major crime that was refusing to provide any information. As we know, it’s not difficult to refrain from saying what we are thinking… but it’s not easy to refrain from thinking when prompted to do so.

There are clearly some privacy and ethical concerns around the use of this kind of technology, but it’s clear that there are important applications for this tech.

And I’ll point out – this only works with fMRI machines right now. These are highly specialized devices that require a ton of power to operate. For that reason, they are only found in lab settings. So we don’t have to worry about generative AI reading our minds as we go about our days.

But it’s not unthinkable that in the near future, smaller electrodes of some kind can be used to capture our brainwaves for the same purpose. This might even be the key to a future brain computer interface that will reduce the need for us to rely so heavily on the old keyboard and mouse.


Jeff Brown
Editor, The Bleeding Edge