DeepSeek model R1, an AI model developed by a chinese startup, skyrocketed on top of Apple Store’s downloads soon after the launch of its latest model on 20th January, surpassing OpenAI’s ChatGPT. This sudden rise has taken every AI investor, especially in the US, by surprise, dropping the stocks of some major tech giants. But, is DeepSeek a genuine breakthrough in the future of AI or just a passing trend? Explore this article to find out.
What is DeepSeek?
DeepSeek is an AI-powered Chinese Chatbot which looks similar to OpenAI’s ChatGPT. It is founded by Chinese stock trading company High-Flyer. The goal of the company is to build AI that could compete with the biggest innovation in the industry like ChatGPT by OpenAI and Gemini by Google. Liang Wenfeng, a Chinese entrepreneur owns both High-Flyer and DeepSeek.
DeepSeek News
DeepSeek Model V3 is capable of solving complex mathematics and can even write coding as effectively as any other product in the market. But OpenAI’s latest o1 model tried to take over DeepSeek’s V3 model, but was overshadowed by DeepSeek’s new launch of its own reasoning model R1.
DeepSeek’s latest reasoning model R1 is competing with Open AI’s o1 model, capable of producing responses similar to how humans enquire problems or ideas. The major highlight of the AI-powered chatbot is that it uses less memory ultimately reducing the overall cost of performing tasks.
DeepSeek vs ChatGPT
DeepSeek’s V3 model is capable of performing tasks equivalent to ChatGPT 4-0 and Anthropic’s Claude 3.5. The bigger breakthrough is not the performance rather the reduced cost attached to it. As per the source, DeepSeek V3 model was trained at a mere cost of $5.58 million USD, much cheaper than ChatGPT-4, which took more than $100 million USD to train itself. (Source).
Moreover, it was trained just by using 2048 H800 GPUs made by NVIDIA. These numbers are way more than major tech giants which are using more than 16,000 H100 chips, a lot more advanced and powerful chip. The claims are mentioned in a research paper released by DeepSeek itself on their official twitter handle.
Reactions across the globe
- Sam Altman, the chief of OpenAI has praised the innovation of R1 model of DeepSeek and has said to deliver much promising models introducing AGI.
- Microsoft has rolled out DeepSeek on Microsoft Azure AI Foundry. The platform has confirmed it from its twitter handle.
- American President Donald Trump call it as a wakeup call for the americans.
How DeepSeek did it
DeepSeek uses an altogether different approach to train its R1 models, working rigorously towards achieving the feat of artificial general intelligence. As outlined in their revealed reserach paper, the company worked on following principles for achieving the desired results:
- Reinforcement Learning: It used large scale reinforced learning approach based on reasoning tasks in order to make the model perform wisely.
- Distillation: It is an efficient learning knowledge transfer techniques from an already developed model to a new developed model.
- Reward Modelling: It is designing an incentive system that guides the AI model during training.
Also, DeepSeek has been released as an open source AI, challenging the major tech giants like Meta, OpenAI and Google operating from the west. This is expected to bring down the cost of AI in the coming future.
What it means
AI models are developed and traded as open source models or proprietary models. Open Source models are cheap and customisable, but require more effort and technical assistance to manage everything. Whereas proprietary models cost more and are easy to manage as the parent company takes care of the maintenance.
Also Read: Suchir Balaji, OpenAI Whistleblower, Found Dead in US Apartment
DeepSeek models are available as an Open Source AI, which means that it can be downloaded and modified as per the user convenience. As most companies require customisation of data, therefore they prefer open source AIs, allowing modifications as per the requirement. Also, technical engineers prefer to access and improve the code on the basis of their requirement.
DeepSeek has disrupted the AI market with the innovation, especially in the west, presenting a cheap open source AI for companies to invest in. Consequently, it has adversely affected the stocks of major tech giants like Meta and NVIDIA. The same concept was briefed by Yann LeCun, Meta’s chief AI scientist as well, saying that Open Source AIs are taking over Closed AIs. Open Source AI gives ample opportunities for creators to use and modify it without asking for permissions.
Security and Privacy
The US government has still not considered DeepSeek fit amid concerns looming over their national security because of the amount of data it captures from the users. It claims of acquiring a users personal data, date of birth, mobile number, phone’s information, keystroke patterns, IP address and storing it in their servers in China. But we have seen in the past that these information are easily shared with advertising partners and corporate groups. Australian science minister, Ed Husic, has supported claims from the west regarding the security concerns that looms over DeepSeek. (Source)
As per renowned researchers, users, especially working on some confidential data need to make themselves aware of the associated risks before actually start using these services. (Source)
The future of technology
The ever-increasing rise of technology is multiplying the chances of it taking control of things over humanity in the future. The process of it has already started. In this scenario, it becomes equally important for us to draw a line where technology should ideally impact us instead of harming. Also, the consequences of data security, which are way more than just definitions, are demanding privacy in exchange of convenience it offers. Let’s learn how we can effectively manage technology, reclaiming control over it, instead of becoming a slave.
Most Effective Use of Technology
The most effective use of technology is not in advancing too much, rather it is figuring out where to draw a line and minimise its consequences simultaneously. And in the absence of a true guide we have misused it to gather sins. Technologies like DeepFake is a prominent example of it. Again, the solution to it lies not in developing too much, rather hidden in embracing the right spiritual wisdom by seeking refuge in a Complete Saint.
At present, Jagatguru Tatvdarshi Sant Rampal Ji Maharaj is the only one Who is highlighting a crucial observation and giving the reason behind the rapid development of technology in the past five decades. It holds a much deeper concept than it appears. The development of this technology is only given to make way for the noble message of Supreme God Kabir to reach every home, fostering peace, humanity, and harmony into the society.
Sant Rampal Ji Maharaj has commenced the golden age where every pious soul can understand the true spiritual knowledge explained by Him by making the most of the available technological assets, primarily the internet and AI (artificial intelligence). And it helps seekers in verifying the impactful proofs manifested by Him to prove the identity of Kabir Saheb as Supreme God as outlined in our sacred texts, and also aids us in adopting the right method of worship to attain inner fulfilment.
All of these references are available on His website www.jagatgururampalji.org and explained precisely in His spiritual discourses on His youtube channel ‘Sant Rampal Ji Maharaj’.
