How DeepSeek could change the landscape in education

Abstract: DeepSeek, a Chinese AI startup, is disrupting the U.S.-dominated AI landscape with an affordable, open-source alternative to models like ChatGPT. Offering top-tier performance at a fraction of the cost, it has the potential to transform education by making AI tools accessible to schools and students globally. However, concerns over censorship, bias, and data privacy remain critical challenges.

Since Generative-AI became a household term, the global AI race has been dominated by U.S. giants like OpenAI, Google, and Microsoft. But a new player is shaking things up: DeepSeek, a Chinese AI startup, is making waves with its innovative and cost-effective approach. While much of the focus has been on its technical achievements, DeepSeek’s potential to transform education is equally groundbreaking.

In this article we will explore DeepSeek’s recent success and why school administrators and educators should be paying attention.
‍

What Makes DeepSeek Different?‍

DeepSeek stands out for two key reasons:
‍

• Affordability
• Transparency 

DeepSeek delivers comparable performance to that of Anthropic's Claude 3.5 or ChatGPT 4o. In Berkeley’s Chatbot Arena LLM, a leaderboard developed by researchers at UC Berkeley, Deepseek is ranked among the best performing models. According to the technical report by Deepseek, their AI outperforms the current leading AI models. (See graph below)

DeepSeek does all this at a fraction of the cost, for reference, this is the API pricing for using DeepSeek Chat vs OpenAI’s ChatGPT. 

This makes it a practical option for schools and students on tight budgets given it’s less than 5% of the cost of ChatGPT. Its open-source design, under “MIT license”, gives users the freedom to customize and improve the platform; a level of flexibility that closed systems like GPT do not offer. Even Meta’s Llama’s 3 which is open source has a custom license with limitations.

OpenAI, despite its naming, has decided to keep their model “closed” and not share it with the open world. Some critics of OpenAI would argue that the real “Open AI” i.e. AI that is truly open is now coming from China.

Impact on the Market

It has been estimated that OpenAI and Anthropic required $100M to 1B to train their models ^[1]. Deepseek has trained their model with a reported $6M budget ^[2]. DeepSeek has also thrived despite hardware limitations. While U.S. companies rely on Nvidia’s most advanced chips, DeepSeek uses older chips due to export restrictions. This proves that innovation doesn’t necessarily require the latest technology. These datapoints have caused the market to believe that instead of more compute power for better models, researchers may need to focus on making them more efficient. This news is why Nvidia stock recently dropped by ~15%, which equates to a $450B loss in its valuation. DeepSeek has now surpassed ChatGPT as the top app on the App Store.
‍

‍

Impact on Education

So, what could DeepSeek mean for education? For students, it could represent equal access to the latest and best technology for everyone, regardless of their financial circumstances. The cost barrier would no longer be a factor limiting accessibility and holding students back. By offering the models that rival, or even surpasses the biggest models such as OpenAI or Claude, at a fraction of the cost, DeepSeek bridges resource gaps, ensuring that financial privilege no longer determines who has access to advanced AI tools.

For instructors and educational institutions, DeepSeek could be a real game-changer thanks to its open-source design. Schools could host the model on their own cloud providers, giving them access to a top tier LLM and have complete control over how it’s used. And because it’s affordable and runs on basic hardware, it’s an option even for schools with tight budgets. Doing so will help them cut on their spending on AI technologies (by 95%) and invest more in students’ needs. This could also open up more opportunities for instructors to integrate AI into their classroom, promoting and facilitating AI literacy without impacting their budget.

Its affordability and scalability make it a viable solution for global education challenges as well. In developing countries, where access to quality educational resources is often limited, DeepSeek could provide great support.

Challenges and Opportunities

Of course, there are challenges to consider. How do we ensure access to a neutral and uncensored platform? As a technology developed in China, DeepSeek has already faced criticism. The model has displayed signs of bias and censorship, for example, declining to provide further details and information on historical events like Tiananmen Square, which are viewed as sensitive topics.

‍

‍

This essentially raises the question on whether the app can be relied on to provide unbiased and educational information.

With safety and privacy being critical concerns in education, DeepSeek could face scrutiny on how it handles sensitive information and protects its users data from cyberattacks, after already falling victim to one^[3]. At the moment, their privacy policies on those matters are still unclear, leaving many questioning its reliability.

What’s Next

DeepSeek could mark the start of a new era in education, one where leading AI technologies are not just a luxury for the few, but a tool for the many. Its affordability, accessibility, and open-source nature could revolutionize how we teach and learn, making high-quality education more equitable and inclusive.