AI Is Improving Its Memory
It's that time of the year again. Summer is over. Vacations have come to an end. And your kids or grandkids are going back to school. You are either back in the swing of things and going back to your normal work routine or are enjoying the silence that is now in your home.
I remember when I was younger and going back to school after summer vacation... I was always anxious about the upcoming year and what it had in store for me. And I was also anxious because it was hard to remember what I'd learned before summer started. I was nervous that everything I'd learned from the year before would be a distant memory and that I'd never remember it.
Human brains need exercise, especially when it comes to retaining and remembering things. Artificial intelligence (AI), or machine learning, works like the human brain but better. Well, it's actually pretty superior to the human brain when it comes to retaining and remembering the information it's learned.
And this is one of the reasons that AI will be a useful tool for our future...
AI's 1 Flaw
AI systems have been pretty remarkable with the amount of information they've been able to retain. But there's always been one flaw to these AI systems that's been hard to overcome. The systems are unable to comprehend the underlying logic of what they do.
The systems don't comprehend anything beyond what they've already been "taught." This is despite how similar the new information might be to previous info. Essentially, you have to train them all over again. Once the algorithm is trained, it's impossible to update again.
Sign up for the Wealth Daily newsletter below to stay on top of the greatest value investment ideas... You'll also get our free report, Seven Strategies for Tech Investing.
This is called catastrophic forgetting. It's the tendency of AI systems to forget what they've learned before whenever they take on new information.
Scientists have been working on this AI flaw for years and looking for a solution. And it's been an uphill battle for most scientists. That is, until recently…
AI Systems Are Improving
During a presentation at the Joint Multi-Conference on Human-Level Artificial Intelligence, Irina Higgins, a senior research scientist at Google DeepMind, announced that her team had begun working on finding the solution to this AI flaw.
Scientists are hyper-focused on AI because it'll be the backbone of the entire world. The potential that AI has for the future seems a little unbelievable. But it could be one of the most realistic and tangible technologies expectations in a long time. We've been waiting for the next tech revolution. It's here, and it's constantly improving...
Higgins has developed an AI agent, similar to a video game character and is controlled by an AI algorithm, in the hopes that the AI agent would be able to think more creatively than a typical AI algorithm could. This would pave the way for a bright future for AI. Higgins says: "We want a machine to learn safe common sense in its exploration so it's not damaging itself."
Higgins' AI system has the ability to learn what an object is, how it relates to that environment, and how it might look from other angles or in different lighting. This helps the algorithm to recognize an object, despite its many angles and different environments. The algorithm is able to identify the differences between what it encounters and what it's seen in the past.
This is a huge step for AI. But of course, it's only the beginning! The AI system is able to update itself and information it's retained about the world. And it can do that without having to be retrained or relearn everything all over again. This will greatly improve the AI systems that we already have in place.
Again, this is a huge step for AI in becoming more than just an algorithm and becoming more like the human brain.
And now is the perfect time to start investing in AI — before this innovative technology really starts to take off.
Learn how to start investing in AI here.
Until next time,
Editorial Director, Park Avenue Digest