Nvidia’s CEO says a simple solution to the AI nightmare could boost crypto — but only if it works
Nvidia CEO Jensen Huang recently said on the record that he believes human-level artificial intelligence (AI) will probably become a reality in the next five years, and the industry's biggest challenge, the illusion, will be easier to solve. .
Hang's comments came during a March 20 speech at Nvidia's GTC developers conference in San Jose, California.
During the event's keynote, Huang proposed Artificial General Intelligence (AGI). According to a report from TechCrunch, the CEO told reporters that the advent of AGI is a matter of benchmarking:
“If we describe AGI as something very different, I believe we'll get to a test set where the software program works very well – or maybe 8% better than most people – within 5 years.”
It is unclear exactly what tests Huang is referring to. In the context of artificial intelligence, the term “general” normally implies that, regardless of the measurement, an average human intelligence can do anything given enough resources.
He also went on to discuss the “illusions” of training large language models to function as generative AI systems. Illusions occur when AI models produce new, often incorrect, information that is not present in the data set.
According to Huang, solving nightmares should be a simple matter. “Add a rule: you have to find the answer for every answer,” he told audience members, “AI shouldn't just answer.” To decide which of the answers is best, one should do research first.
Putting aside Microsoft's Copilot AI, Google Gemini, OpenAI's ChatGPT, and Anthropoc Cloud 3's ability to source all their results from the Internet, if the AI nightmare problem is solved once and for all, it could transform a myriad of industries. including finance and crypto.
Currently, the creators of the aforementioned systems recommend caution when using generative AI systems for tasks where accuracy is important. ChatGPT's user interface, for example, warns that “ChatGPT may be buggy” and advises users to “consider checking important information.”
In the world of finance and cryptocurrency, accuracy can mean the difference between profit and loss. This means that generative AI systems currently have limited functionality for financial and crypto professionals.
Related: CFTC Warns AI Can't Pick Your Next Crypto Winner
While there are experiments involving trading bots powered by generative AI systems, they are usually tightly regulated to prevent autonomous execution – meaning they are designed to execute trades in a strictly regulated manner, similar to placing limit orders.
Generative AI models should be able to trade and make financial recommendations and decisions completely without human input, ostensibly, if they don't suffer from churning out completely artificial results. In other words, if the fantasy problem in AI is solved, fully automated marketing can become a reality.