12 hours ago
9
January 30, 2025 | 12:11 pm
TEMPO.CO, Jakarta - The emergence of Chinese startup DeepSeek has wreaked havoc amidst the tech industry, threatening the dominance of AI pioneers like Nvidia. Since its first launch, the free AI assistant continues to gain traction among users and investors alike. As seen on Apple’s app store, the Chinese AI models are taking over their rival ChatGPT in downloads.
Reuters reported that DeepSeek’s AI assistant uses less data at a fraction of the cost of incumbent services, attracting attention in global AI circles. Setting a record one-day loss for Wall Street’s company, Nvidia, of $593 billion, here’s everything you need to know about the Chinese startup DeepSeek as highlighted by various sources.
What is DeepSeek?
DeepSeek is a Chinese startup artificial intelligence (AI) company founded by a graduate of Zhejiang University, Liang Wengfeng, in July 2023. Previously, Liang’s fund made an announcement on its official WeChat account that it was starting over, focusing on the creation of a “new and independent research group, to explore the essence of AGI,” as per Reuters.
AGI, also known as Artificial General Intelligence, is an autonomous system surpassing humans in most economically valuable tasks, according to OpenAI.
Citing CBS News, Liang reportedly has around $8 billion in assets and had bought a “stockpile of Nvidia A100 chips,” which became the basis of DeepSeek. As a result of utilizing a type of tech that is now banned from export to China, the AI company has taken the low-cost approach to work with limits to the number of available computer chips.
DeepSeek R1 is currently available for anyone to access, use, modify, and share without a restriction of proprietary licenses.
Is DeepSeek Efficient?
Forbes noted that the DeepSeek R1 model is most similar to OpenAI’s o1 technology, with both having the specialization in turning requests into a chain of logical thoughts. Unlike OpenAI’s ChatGPT, DeepSeek R1 isn’t designed to operate as a Q&A chatbot.
DeepSeek’s R1 model requires less memory use, resulting in the reduction of computational costs. Notably, the artificial intelligence bot has received a nod from researchers for its capability to understand complex reasoning tasks in mathematics and coding, The Guardian reported.
On a recent MATH-500 test, DeepSeek’s R1 has achieved a glamorous performance that is on par with o1 in multiple benchmarks. As a result, Chatbot Arena ranks R1 as the third-best AI model in modern dates.
What Are DeepSeek Capabilities?
Since the day DeepSeek’s R1 was released to the public, the AI model has been garnering notice. Here are a few capabilities the computing system can do:
1. Ideas The
The DeepSeek R1 model is designed to answer questions, solve logic problems, and write its own computer programs. Similarly to OpenAI’s ChatGPT, users are able to use DeepSeek to generate brainstorming ideas.
2. Writing
Although there have been numerous articles highlighting the cheaper costs of DeepSeek’s AI models, it doesn’t put restrictions on users as much. The chatbot machine can write a paragraph with over 100 words describing a specific idea.
DeepSeek’s rapid rise in the AI industry is challenging established giants like Nvidia and OpenAI. With its low-cost, high-efficiency model and open-access approach, the Chinese startup is proving to be a formidable player in the race toward advanced AI capabilities.
FORBES | REUTERS | THE GUARDIAN | CBS NEWS | NEW YORK TIMES
Editor’s Choice: Top AI Companies to Invest in: NVIDIA Remains the Top Pick
Click here to get the latest news updates from Tempo on Google News
Donald Trump Says DeepSeek's AI Should Be a 'Wake-up Call' to U.S. Industry
1 hari lalu
Donald Trump said that Chinese startup DeepSeek's technology should act as a spur for American companies.
Alibaba Cloud Sees Indonesia as Key Market for Cloud, AI Technology in Southeast Asia
7 hari lalu
Alibaba Cloud is committed to supporting the development of Indonesian tech talent, particularly in the AI domain.
A Brief Look at Stargate, the US$500bn Investment in AI Infrastructure
7 hari lalu
Donald Trump announced a joint venture called Stargate intending to invest US$500 billion in AI infrastructure. Here's a closer look at Stargate.
Microsoft Tests AI-Based Search Feature in Windows, Here's How It's Used
9 hari lalu
Microsoft is reportedly testing an artificial intelligence (AI)-based Windows search feature in a new dev channel for Windows 11 Insider testers.
6 Passive Income Ideas with ChatGPT Prompts
17 hari lalu
This article introduces top passive income ideas with ChatGPT prompts, in which you can gain greater profitability & streamline your workflow.
5 Safest Jobs from AI: Standing Strong Against Automation
21 hari lalu
In the wake of today's rapid advancements in artificial intelligence, these safest jobs from AI stand out as a reassurance for people affected.
This Year, Microsoft Plans to Invest US$80 Billion in AI Data Center Development
24 hari lalu
Microsoft plans to invest about US$80 billion in fiscal 2025 to expand data centers used to train artificial intelligence (AI) models.
Constitutional Court Rules Against Excessive Use of AI-Manipulated Images for Election Campaign
27 hari lalu
Constitutional Court said false information generated by excessive image manipulation through AI can damage voters' ability to make quality decisions.
Top AI Companies to Invest in: NVIDIA Remains the Top Pick
35 hari lalu
The rapid evolution of artificial intelligence (AI) has shifted the attention to numerous top AI companies to invest in. See the outlook here.
Indonesia's Komdigi Ministry Preparing to Regulate AI Use
36 hari lalu
Deputy Minister of Komdigi said that the regulations of AI use will include principles such as transparency, accountability, and humanity.
