Grok, Elon Musk's AI chatbot, to release a revised version, stepping up competition with ChatGPT.
An enhanced version of Grok 1.5 has been released by Elon Musk's AI chatbot Grok, which is stepping up competition with OpenAI's ChatGPT.
Grok's enhanced version has a context length of 128,000 tokens and enhanced reasoning. Because of this, Grok can now store information from much lengthier documents in its memory, up to 16 times longer than its prior context.
In the next few days, early testers and current Grok users on X will be able to access Grok.
The business announced in a blog post that an updated version of Grok was now available.
“Introducing Grok-1.5, our latest model capable of long context understanding and advanced reasoning. Grok-1.5 will be available to our early testers and existing Grok users on the X platform in the coming days. By releasing the model weights and network architecture of Grok-1 two weeks ago, we presented a glimpse into the progress xAI had made up until last November. Since then, we have improved reasoning and problem-solving capabilities in our latest model, Grok-1.5.”
“One of the most notable improvements in Grok-1.5 is its performance in coding and math-related tasks. In our tests, Grok-1.5 achieved a 50.6% score on the MATH benchmark and a 90% score on the GSM8k benchmark, two math benchmarks covering a wide range of grade school to high school competition problems. Additionally, it scored 74.1% on the HumanEval benchmark, which evaluates code generation and problem-solving abilities.”
Grok's ability to answer questions regarding politics that other generative AI chatbots usually avoid answering, such as conspiracies and more contentious political theories, is one of its most notable features. The X platform gives the chatbot access to real-time information, which enhances its personality. It responds to acrid queries that other AI systems might ignore.
Elon Musk made a clear swipe at OpenAI's ChatGPT by stating that one of the objectives for his AI chatbot Grok, which competes with it, is to be the funniest chatbot. This was mentioned in response to criticism levelled at other chatbots for being overly sensitive.
Musk has strongly criticized ChatGPT for what he sees as censoring some replies to specific questions. This, he claims, is what inspired Grok to be launched as a more accurate portrayal of reality.
In an X Spaces talk, the billionaire owner of X (previously Twitter) discussed the future of Grok with investor Cathie Wood.
He said,
“Grok is pretty funny. I think one of our goals for Grok is to be the funniest AI. So I mean if you ask Grok to provide a vulgar roast, it’s really good.”
In his conversation with Wood, Musk added that Grok will be a significant truth improver.
“That is our goal, to be maximally truth-seeking, maximally curious, minimizing the error between perceived reality, and described reality, and actual reality,” he added.
In March 2023, Musk formed xAI, which produced Grok, which was made available to the general public in November. The generative AI chatbot from xAI, Grok, is built on a sizable language model. Its core algorithm leverages deep learning and analyzes a substantial volume of data to produce content.
Grok's AI chatbot is getting an update, almost two weeks after Elon Musk's xAI business made the AI model behind it publicly available.
With the release of Grok-1.5 on Thursday, the business asserted that its most recent model could comprehend longer documents, respond to more intricate instructions, and engage in more sophisticated reasoning.
Although Grok-1.5's coding and math abilities seem to have improved from the original 1.0, its announcement post reveals that it still performs better than OpenAI on one important HumanEval test but falls short of Google's Gemini Pro 1.5 AI, Anthropic's Claude 3 Opus, and OpenAI on some benchmark tests.
Similar: Get to Know Grok: Elon Musk Presents "Spicy," a "Humorous" and "Sarcastic" AI Chatbot
On the HumanEval benchmark, which consists of 164 difficult programming problems that are not part of the AI model's training set, Grok-1.5 performed better than GPT-4. Grok-1.5 scored 74.1%, Gemini Pro 1.5 scored 71.9%, and GPT-4 scored 67%.
Grok-1.5 did nearly as well as Google Gemini, scoring 83.7% on the MMLU test, which spans knowledge of 57 subjects from an elementary to an advanced level. Grok-1.5's score was 81.3%.
It also received a score of 50.6% on the MATH exam, a benchmark that includes issues from elementary school math competitions to high school arithmetic tests, which is quite similar to the GPT-4's 52.9% result.
In a social media post on Friday, Musk said that Grok 1.5 will be out by next week on X, which used to be Twitter.
The proprietor of X has great hopes for the upcoming version of Grok; in his words, "On all metrics, the next step beyond Grok-1.5 will outperform the AI currently available."In training now," Grok 2 is," he stated in the post.
Currently, Grok AI is only accessible to users of X with a Premium+ subscription that costs at least sixteen dollars a month.
Musk sued OpenAI, a rival to xAI, earlier this month, requesting a court decision compelling OpenAI to disclose the technology and research supporting its artificial intelligence.
Challenges to the Status Quo
The choice made by Elon Musk to open-source Grok raises concerns about AI's future and criticizes accepted practices in AI development.
Here's a comparison between Grok and ChatGPT, even with its impending open-source release:
Data and Access:
- Grok: keeps information current by concentrating on real-time data and providing access to the X platform (previously Twitter). Beta testers will have limited access, however later on, paid X users will get more access.
- ChatGPT: trained using a sizable, static corpus of material from the public internet. extensively available in both free and paid tiers.
Strengths:
- Grok: X platform integration aims for informative, even amusing responses and may provide more up-to-date information.
- ChatGPT: shines in its versatility, using a large amount of training data to handle a wider range of stimuli and produce original text structures.
Weaknesses:
- Grok: Currently has restricted access and is primarily dependent on one platform (X) for data.
- ChatGPT: potentially lacking in real-time updates and susceptible to bias in its training set.
Seeking Funding for Grok’s Development
Apart from the announcement of going open-source, Musk has been actively pursuing funds to advance the development of Grok.
According to reports, he has made contact with Middle Eastern sovereign wealth institutions and affluent Hong Kong families to raise some $6,000 million for xAI.
Musk's commitment to pushing the limits of AI capabilities and looking into new areas for innovation is demonstrated by this funding campaign.
Conclusion
The decision by Elon Musk to release Grok as open-source is a critical turning point in the advancement of AI technology. By highlighting transparency, accessibility, and cooperation as crucial elements of responsible AI research, the action subverts accepted conventions.
Musk's vision for the future of AI technology sets the foundation for a new era of innovation and ethical considerations in AI, while the tech community awaits further information about the open-source release.
REFERENCES
https://www.tekedia.com/elon-musk-owned-ai-chatbot-grok-set-to-launch-an-upgraded-model/
