What is Deep Seek AI?

China has taken over the world by storm since it launched its cost-effective A.I model called deep seek AI, is also on the top of the list of downloads among all the applications on the Apple application store. It has also sunken the stocks of companies like Nvidia. Almost 1 trillion dollars were lost in the stock market because of this AI rival that stands firm against its competitor of Open AI chat GPT. This has also been called the Sputnik moment in terms of A.I rivalry, lastly it was the war to go to space at the time of Cold War, now the war is transformed into developing the most powerful and cost-effective A.I chips and pre-trained transformers. Chinese have taken the lead since their version of A.I is 10 times cheaper and as effective as the new model of Chat GPT, not only this it is free to use also. Trump said while addressing reporters recently that this is a wake up call for the companies in silicon valley, therefore we must be focusing on winning this competition from china and other rivals. 

What makes Deep Seek AI so special?

The company behind Deep Seek AI has claimed that this AI model has taken a fraction of what has been invested to build Chat GPT. And only a few advanced chips have been used to develop this Deep Seek AI model. The chip-making giant Nvidia shed at least 600 billion dollars and at least 482 billion pounds last monday, that’s one of the biggest losses in the history of the United States. Now, U.S also wants to ban the exports of one of the advanced ships to China, which can slow down the process of AI development in China, but there are other ways to compete with China also, that would create friendly competition among the two countries. President Xi Jinping of China has also doubled down his stance on AI development, declaring it as one of the top priorities of the nation. This is a huge transition from the Chinese competitors since they have moved on from manufacturing of traditional goods and now are taking huge strides in developing cutting-edge technology like; integrated circuits, electric vehicles and now AI. 

What is Artificial intelligence and how does it work?

Current Artificial intelligence models are transformers, let’s take an example of a generator that we use as a power backup. So, that generator is basically a transformer which converts mechanical energy to electrical energy. Similarly, Current AI models are being fed a lot of data from the internet to train them to become intelligent, for instance, an AI should know the basics of mathematics, which means AI’s brain is eating all of the information that is related to mathematics, then to understand science and physics, you have to feed a lot of data including books and papers. Therefore, AI is fed a lot of cleaned data and then it is programmed to make sense of it and respond whenever a user sort of gives any prompt as input. 

So, the current form of these artificial intelligence models interacts with human beings like chatbots, and in future, the same artificial intelligence models could take a physical form. Now, these chat bots have become multi-modal, meaning that user cannot only give text as input, but also able to upload voice images to receive their outputs or results. These GPT’s can also be customised through various plugins, for instance; if you are looking to create speech-to-text then you can find a specific plugin to do that in the chat module. Similarly to generate images like; Wall-E is a separate platform to generate images but you can do the same thing in ChatGPT. Chat GPT has become one of the vital tools when it comes to studying also, it’s not just a search engine, it can be a teacher that you didn’t had access to, therefore you can learn anything about it, whether it is physics, mathematics, electronics, computer science etc. 

Why DeepSeek?

DeepSeek is the name of a free AI-powered chatbot, which is developed by a chinese company and it has become the biggest rival to ChatGPT by OpenAI. DeepSeek AI feels more human than ChatGPT because it will show all of its thought processes which are typically done by human beings.  For instance; Ohh, I forgot to do that, oh, that should be done if I have to solve that too. For things like that exclamation mark situations, Deep Seek AI would show you and write all of its thought process including what it is doing what it is doing. DeepSeek has beaten the ChatGPT optimised 4.0 model based on its English, coding and mathematical abilities. O1 is OpenAI current model and R1 is DeepSeek’s reasoning model the current model R1 of DeepSeek has less memory than its rivals, meaning it uses less cost to perform tasks that are even technical and can only be performed through other models by paying a premium. 

The current model also provides accurate sources rather than blocking any sort of information since chinese government censorship is also a huge challenge for its AI aspirations but this model cannot be used as a use case for that. It has been said that the cost of making this AI is around $6m (£4.8m) to train, a fraction of the “over $100m”. It was also noted that the founder of DeepSeek built up a store of Nvidia A100 chips and these chips have been banned from being exported to china since September of 2022. This collection of chips is around 50,000 believed by some, that led DeepSeek AI to build a powerful AI model, additionally, these chips were paired with some cheaper chips to make it more sophisticated. 

Who is behind DeepSeek?

Liang Wenfeng is the founder of DeepSeek, it was founded in December 2023, and released its first large language model in the coming year. Liang Wenfeng graduated from Zhejiang University, and has a degree in electronics engineering. He was recently seen in a meeting hosted by China’s premier Li Qiang, reflecting DeepSeek’s vision and how his background is different from the American AI entrepreneurs who are from Silicon Valley. He also holds knowledge of finance along with a degree in electronics and computer science. He also stated that China cannot remain a follower anymore when it comes to AI, if America can develop high-quality AI models then China can do so too.  He also stated that there is always a two-year gap between Chinese and American products, he said that if that gap is not being reduced then China will always be a follower in this AI race. 

Final Verdict is that in the days to come China is following up closely with its U.S counterparts when it comes to the development of cutting-edge Artificial intelligence models. But still U.S has the upper hand since all of the advanced chips are being developed in U.S. And China is very much limited in this case to utilise that sort of technology and then develop and advance its own AI capabilities. Only time will tell how this gap will be reduced and how this competition will be unfolded among these two giants.