AI’s Biggest Flaw Solved?

Written By Luke Sweeney

Posted February 21, 2023

If you haven't been able to try your hand at AI programs like ChatGPT and DALL-E, you're not alone. The endless wait lists mean only a select few can access them.  

It’s the same reason nobody drives in New York — too much traffic. 

These wait lists are restrictive partially because the programs are still in the early development stages, but there’s another reason they haven’t already become an app on your phone: They use too much power. 

The computational requirements for these programs are enormous. That’s why only major players like OpenAI, Google, and Microsoft have been dominating the headlines. 

And even if you had a world-class computer system, the training data that feeds the AI’s brain comes from Google or Microsoft 9 times out of 10.

It’s almost painful for me to admit, but this new era of generative AI absolutely depends on Big Tech. 

However, that doesn't mean your only investment choices are on the top shelf. AI is flexible enough to be specifically tailored to almost any industry. 

Google and Microsoft can’t possibly have expertise in every area — though it won't stop them from trying. 

They can provide data and maybe a one-size-fits-all platform, but I doubt they'll spend time building a program customized for utility management, or one that calculates the best cooking time for a bakery. 

This isn't a discovery that will only benefit Big Tech. It’s a major equalizer for small businesses and regular people too. 

The Future of AI Belongs to Everyone

AI is trained by data — and, in a way, so are humans. 

AI couldn't exist without all of us constantly engaging with each other over the internet. If you've ever made a post on the internet, there’s a good chance a tiny part of you is coded into every modern AI program. 

That adds up to a lot of data. Experts estimate that we create somewhere around 650 petabytes each year. 

To put that in perspective, 1 petabyte is equal to 1 million gigabytes. Computing that incomprehensible amount of data consumes roughly 10% of the world’s electricity, or over 100 billion kilowatt-hours per year. 

Now, that power usage estimate was made years before AI became commonplace. A third-party analysis recently showed that ChatGPT consumed 1,287 megawatt-hours (MWh) and led to emissions of more than 550 tonnes of carbon dioxide. 

For comparison, that’s the same amount as a single person taking 550 flights back and forth between New York and San Francisco. 

Other experts claim that training a single AI model will emit as much carbon as five cars would throughout their lifetime. 

It’s a problem that nobody is talking about so far. Just like the inevitable EV takeover, the transition to AI-powered programs will require a massive infrastructure overhaul. 

Personal computers will need more memory and better thermal control — either that, or everything will be offloaded to a cloud server. 

But what about the data centers hosting those servers? They'll need an exponential increase in power to handle millions or even billions of AI-enhanced user requests. 

It’s a problem the world has so far ignored as we play around with these early beta tests and if it continues to be ignored, it could push the full-scale AI rollout back years. 

If only there were a magic solution…

Luckily, AI Has Solutions for Its Own Problems

Perhaps the biggest benefit of AI is its ability to optimize. It can take in every shred of operational data for a business and figure out where to improve. 

For data-heavy tasks like determining which route a delivery truck should take, it’s almost indispensable. 

One such task that AI is well suited for is improving the efficiency of things like data centers. That’s right, AI can even work to improve itself.

Among other things, AI can monitor servers, analyze network congestion, and optimize disk utilization to predict data outages and eliminate downtime. 

Major data centers are notorious for falling victim to attacks — physical as well as digital — which can cost service providers huge amounts of money. AI can scout for vulnerabilities and prevent these attacks from ever occurring. 

That’s just the tip of the iceberg. I could go on all day talking about the endless potential use cases for AI, but it’s much easier to just show you. 

Our team is working to bring you a comprehensive investment guide to take advantage of this AI boom. Our complete research compendium will be available very soon. 

For now, take a look at some of the industries AI can revolutionize today:

  • Nuclear Fuel: Advanced AI software can simulate the dangerous conditions inside a reactor, allowing engineers to test new designs without risking another Chernobyl.
  • Quantum Computing: Building a functional quantum computer that can rival typical supercomputers is one of the most complex tasks in the field of physics. AI can use its brainpower to find ideal materials or designs.
  • Green Energy: Transitioning to green energy is as much a logistics problem as it is an engineering one. While an AI could definitely help think up a new green fuel, one of its more immediate uses is to optimize the electrical grid for renewable power.

Stay tuned for more updates on this thrillingly lucrative industry.