Connect with us

Hi, what are you looking for?

Economy

Elon Musk’s new chatbot, Grok, scored 59 percent on a Hungarian math test

Elon Musk’s new chatbot, Grok, scored 59 percent on a Hungarian math test

xAI, Elon Musk’s latest company that deals with artificial intelligence, has announced that its chatbot called Grok is ready. The system was developed within two months, but according to indications, it is capable of competing with well-known competitors.

the Read an article on Wired It reminds him: Last week Musk participated in a conference in Bletchley Park near London, where politicians, businessmen and researchers discussed the risks of artificial intelligence. However, he has now proudly announced his new “less restrictive” regime. “It also answers interesting questions that other AI systems dismiss,” Grock says. Advertisement XAI, which also highlights the program’s “rebellious features.” Over the past year, OpenAI, Microsoft, and Google have been constantly talking about “responsible” AI, highlighting how they can prevent criminals or other malicious users from abusing their systems.

“Don’t use it if you hate humor.”

– On the other hand, Grok’s future interlocutors at XAI were warned. The program cannot be tried at this time. Those who are curious can sign up for a waiting list, and if Elon Musk’s promise is true, users who pay for X, Twitter’s successor, will be the first to try it. However, they have already shared some select information about what the system is capable of. Math, general literacy, and programming tasks were also given to the Grok-1 model, which generally performed better than GPT-3.5 running under ChatGPT, but worse than the more advanced GPT-4. What’s particularly interesting is that the developers, including Hungarian Christian Szegedy, have also made the AI ​​available for local mathematics graduation assignments for 2023. The program scored fifty-nine percent, compared to sixty-eight percent for GPT-4.

Your puppy has been trained on thirty-three billion parameters, meaning it has learned based on this large amount of data. For GPT-3.5 it was 175 billion, for GPT-4 it was 175 trillion. According to the expert interviewed by Wired, the program’s performance and numbers seem realistic, and the chatbot’s answers will be supplemented by those written on

What exactly it means that the system gives “hot” answers, even to questions that other platforms cannot answer, will quickly become clear during public tests. However, according to the creators, Grok will improve from week to week, so maybe one day he will get an A in math.

(Cover image: Jaap Arens/Noor Photo via Getty Images)


Arithmetic, Geometry and Physics lessons have already been solved by Google
For all this, the search engine and the Google Lens app can interpret images, videos, and text at the same time.


Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

You May Also Like

Top News

In a harrowing incident that has shaken the community of Lewiston, Maine, a series of shootings on Wednesday evening resulted in a tragic loss...

Top News

President Joe Biden’s abrupt departure from a speech on the U.S. economy at the White House on Monday sent a ripple of speculation and...

World

Chinese scientists have discovered a little-known type of ore containing a rare earth metal highly sought after for its superconducting properties. The ore, called...

Tech

A dangerous application appeared in the Apple App Store disguised as a known program. 24.hu reported the Based on TechCrunch article. Dangerous app in...

Copyright © 2024 Campus Lately.