• A second front-runner for small modular reactors
  • Need a medical diagnosis? Ask the AI…
  • A new tool for detecting AI-generated content

Dear Reader,

Every time I see proclamations about some new green energy breakthrough in decarbonization for a manufacturing process, I’m genuinely hoping the news is real.

Sadly, I’m almost always disappointed.

The latest proclamation centers around a $120 million venture capital round for a private steel manufacturing company – Boston Metal – that claims to completely decarbonize the steel manufacturing process.

Sounds great, right?

About two billion tons of steel are produced annually and account for about 10% of global carbon emissions. Obviously, if global steel production could be completely free of carbon emissions, it would be fantastic.

Today, steel is typically made using metallurgical coal. Metallurgical coal is also referred to as coking coal. It’s heated to 1,000 degrees Celsius and then added into a blast furnace with iron ore. It’s then heated to more than 2,000 degrees Celsius. That intense heat converts the iron ore into a liquid – molten iron – which is eventually mixed with alloys into forms of steel.

However, Boston Metal claims to have developed a “green” process using electrolysis that doesn’t require the use of coal. Boston Metal’s furnace is kind of like a battery with an anode in an electrolyte into which iron ore is deposited. There is also a cathode, just like a battery, and the electricity is used to heat the mixture, which is how the molten iron is produced.

Boston Metal’s Electrolysis Process for Producing Steel

Source: Boston Metal

Boston Metal’s claim of decarbonization stems from the fact that metallurgical coal is not used – only electricity – and so it’s a “green” process.

To be clear, this process uses immense amounts of electricity to produce the liquid iron, which is refined into steel products. So that begs the question… Where does the electricity come from?

The manufacturing plant is in Woburn, MA, just north of Boston. I checked with the U.S. Energy Information Administration, and about 70% of the electricity production for the state of Massachusetts comes from carbon-based energy sources, specifically natural gas. And almost all the actual power produced to feed the greater Boston area is from natural gas.

It is possible for a company like Boston Metal to purchase carbon offsets or carbon credits to be “carbon neutral”, but we should be very clear about how the iron/steel is being manufactured. The plant is fueled by fossil fuels – in this case natural gas. Natural gas is burned to produce the electricity required for Boston Metal’s “green” iron/steel manufacturing process.

Not so clean and green after all…

Just because a process uses electricity doesn’t mean it’s clean. In this case, it’s just a displacement of the physical location where fossil fuels are burned and where carbon emissions take place. And the worst part is that anywhere between 7–15% of electricity is lost through transmission. That means power plants need to burn more fossil fuels to account for transmission loss.

This is why nuclear fission, or preferably nuclear fusion, technologies are so critically important. If we radically change how we produce baseload power capable of running 24/7/365 and ensure that 100% of our electricity grid is clean, then we can claim a green or decarbonized process. 

GE Hitachi throws the gauntlet down for NuScale…

Just last week we had a look at one of the most significant developments in the nuclear fission industry in a couple decades. It came when the U.S. Nuclear Regulatory Commission (NRC) certified NuScale’s small modular reactor (SMR) design.

As a reminder, SMRs have many benefits compared to traditional fission reactors. For one, they’re smaller. One SMR reactor could fit on the back of a semi-trailer. That makes them cheaper to build. And each SMR has the potential to power 45,000 homes.

That means they can be centrally manufactured and then shipped to their final destination for installation. This reduces costs and accelerates construction times significantly.

This was the first SMR design to ever be certified. And it paves the way for NuScale to open its own SMR-based power plant in Idaho by 2029.

Well, it didn’t take long for the gauntlet to drop…

A consortium of four companies, led by GE Hitachi, just announced plans to build what they believe will be North America’s first SMR-based power plant. They are targeting 2028 for the grand opening in an effort to beat NuScale to the punch.

The plant will be located at the Darlington New Nuclear Project site. It’s in Ontario, Canada. And the NRC has already announced that they’ll team up with Canada’s Nuclear Safety Commission to certify the consortium’s SMR design.

So the race is on.

NRC’s certification of NuScale’s design last week appears to be a major catalyst for the next generation of nuclear power production.

As a result, I’m sure we’ll see an increase in venture capital (VC) investment in this space in the months to come. There are several U.S. based companies like TerraPower, Holtec, Kairos, and X-Energy that are also developing SMRs that I expect will receive more attention.

I sincerely hope we’re at the start of a massive trend here.

We’ve talked before about how nuclear fission is the best form of clean energy available to us today. There is radioactive waste to deal with… but it’s not difficult to handle if proper protocols are followed. Other than that, nuclear fission doesn’t produce any pollution or carbon emissions. And most don’t know this, but nuclear power is literally the safest form of energy production. Nuclear power is right on par with solar power, and even better than wind, in terms of death rates per unit of electricity produced.

And smaller SMR designs could enable a far more decentralized power grid. We could have smaller power plants servicing a handful of neighborhoods rather than the very large physical footprints of power plants today. That would make our power grid far more resilient than it is today.

We all stand to benefit from SMR technology. And for investors, there are certainly opportunities with this renewed interest in nuclear fission technology.

For any subscribers of The Near Future Report, I’d encourage you to catch up right here to learn about my preferred way to profit from the boom in nuclear power I expect to unfold in the years ahead.

Generative AI is now capable of a differential diagnosis…

Here’s another trend that’s moving fast…

We noticed last week that OpenAI’s ChatGPT passed all three parts of the U.S. Medical Licensing Exam (USMLE). Even more impressive, ChatGPT did this with no specific medical training whatsoever.

For the sake of new readers, ChatGPT is an incredibly powerful generative AI.

ChatGPT can answer basic questions and write essays about any given topic. It can also compare and contrast different philosophies and opinions. And the AI can even write software code upon command.

And it turns out ChatGPT is also a medical whiz. After it passed the USMLE, I suggested that the AI could be used to help doctors make accurate diagnoses in the future.

Well, it turns out the future is already here. A small early stage company out of San Francisco just launched a feature that applies generative AI to the medical field.

The company is called Glass Health. I asked my team to try out its platform. Turns out it’s quite impressive.

To test it out, we entered a hypothetical situation: 54-year old male presenting with chest pain, headaches, body mass index of 25, otherwise healthy. Then we pressed the “Generate DDx” button. DDX stands for “differential diagnosis.”

It took the AI a few seconds to make its diagnosis… but it’s quite thorough. Here’s a look:

Source: Glass Health

After reading through the diagnosis, we clicked the “Generate Clinical Plan” button. And here’s what the AI suggested:

As we can see, the AI suggests doing an EKG, a chest x-ray, a CT angiogram, and possibly even blood work. It also suggests a few prescriptions that could help the patient, starting with Aspirin.

What a practical tool. Most incredible is how quickly it produced useful and actionable information.

And here’s the best part – generative AIs can incorporate the entire body of research in every area of medicine into their diagnoses. No single physician could possibly do that.

What’s more, these AIs can be updated every day. That way they will always incorporate the latest peer-reviewed research into their suggestions.

Now let’s take it a step further…

Suppose we can upload the patient’s DNA sequencing data into the AI as well. Now the AI can take the patient’s specific genetic structure into account to make its recommendations.

What an incredibly powerful tool for precision medicine. I imagine every doctor and nurse out there would find this useful in their daily work.

And as we discussed before, there are areas in the world where there simply aren’t enough doctors to service the population. AI-based tools like this would be a game changer in those places as well.

Plus, these AIs are available 24/7. They never get tired, they never get upset, and they always perform their task with the same level of performance.

Just imagine where this technology will be in just a year’s time.

Generative AI detection tools are proliferating…

Speaking of generative AI – OpenAI just released a tool that helps determine whether something was written by an AI or a human. They are calling it an AI text classifier.

We looked at a similar tool called GPTZero just a few weeks ago. It was designed by a student at Princeton to identify text written by generative AI.

So here we have OpenAI coming out with the same functionality. And it makes perfect sense. After all, OpenAI created the groundbreaking ChatGPT that’s had the industry buzzing the last few months.

The thing is – the tool is far from perfect.

In testing, it’s only been able to definitively identify AI-written text about 26% of the time. And it produced false positives 9% of the time. That’s where it labeled human text as likely AI-written.

Clearly the tool must improve its accuracy before it can be truly useful.

And get this – OpenAI’s AI classifier is just one of a handful of similar tools that have come out since the GPTZero release.

Stanford has developed its own AI classifier tool called DetectGPT. There’s another one called AI Detector Pro. Then MIT’s Watson AI Lab came out with its own tool. And there are several others that have surfaced as well.

It’s amazing that all of these tools popped up in response to ChatGPT. I don’t think I’ve ever seen a faster response to a new technology. That speaks to just how powerful generative AI is.

So this continues to be the most exciting technology of 2023. And it’s having massive investment implications already.

In fact, one of our small-cap AI stocks in Exponential Tech Investor has shot up roughly 90% just in the last few weeks on the back of the generative AI trend.

And this is just the beginning.

Generative AI will be a huge multi-year trend. We’ll have plenty of opportunities to profit as this technology is employed to solve real world problems.

Regards,

Jeff Brown
Editor, The Bleeding Edge