DeepSeek vs. ChatGPT The Open-Source AI Disrupting EdTech and Language Learning
How DeepSeek’s Open-Source Power, Cost Efficiency, and Multilingual Precision could redefine AI in Education.
DeepSeek vs. ChatGPT: A Disruption in EdTech and Language Learning
To say there is competition in the AI sector is putting things incredibly lightly. Since ChatGPT burst into our lives, the AI landscape has been evolving rapidly, and the competition between models has been dominated by American companies like OpenAI and Anthropic. This competition heated up over the weekend with a new Chinese entry in the AI race. DeepSeek is emerging as a formidable contender that for many reasons could be a significant disruptor. But what makes DeepSeek stand out, and how could it disrupt the AI EdTech sphere? There’s going to be a lot to cover, so let’s get into it.
The Battle of Interfaces and Features
One of the first things users notice is the difference in user experience. ChatGPT’s web interface is polished, feature-rich, and intuitive. DeepSeek’s web interface, on the other hand, is barebones and lacks the bells and whistles that make ChatGPT so appealing. However, where DeepSeek shines is in its performance and cost-effectiveness.
For instance, when tasked with reading and analysing a rental contract in Spanish, DeepSeek outperformed ChatGPT, delivering spot-on accuracy. This could be a game-changer for language learning and assessment tools, where precision and cultural nuance are critical. Imagine an AI-powered language app that can not only assess a student’s grammar and vocabulary but also provide detailed feedback on complex legal or technical documents in multiple languages.
The Cost Factor: A Disruptive Force
One of DeepSeek’s most compelling advantages is its cost structure. Unlike ChatGPT, which requires a subscription, DeepSeek is fully open-source and free to use. This opens up a world of possibilities for EdTech startups and app developers who want to integrate AI into their platforms without breaking the bank.
For example, consider an AI-powered language assessment app. By leveraging DeepSeek’s API, developers could create a low-cost, high-quality tool that rivals expensive proprietary solutions. Students could practice reading, writing, and even legal or technical comprehension in multiple languages, all powered by an AI that’s both accurate and affordable.
The Vision Gap (Literally)
One area where DeepSeek currently falls short is image recognition. While ChatGPT can analyse and interpret images, DeepSeek is limited to OCR (optical character recognition). This is a significant drawback for users who rely on visual inputs, such as students studying diagrams, charts, or handwritten notes.
However, this gap could be bridged by combining DeepSeek with other tools. For instance, using OpenWebUI to switch between DeepSeek and a vision-capable model like GPT-4 could provide the best of both worlds. This modular approach could revolutionise EdTech, allowing educators to tailor AI tools to specific learning objectives.
The Identity Crisis: A Quirk or a Feature?
An amusing quirk of DeepSeek is its tendency to “believe” it’s ChatGPT or even Claude by Anthropic when asked about its origins. This identity crisis highlights the model’s training process, which seems to have borrowed heavily from existing AI systems. Despite this, DeepSeek’s performance rivals that of OpenAI’s more expensive models, making it a compelling alternative.
In the context of EdTech, this adaptability could be a strength. For instance, a language learning app could use DeepSeek to simulate conversations with different “personalities” or cultural perspectives, providing students with a more immersive and diverse learning experience.
Running AI on a Phone: The Future of EdTech?
Perhaps the most exciting aspect of DeepSeek is its efficiency. The distilled version of the model can run independently on a phone, making AI-powered education accessible to anyone with a smartphone. Let’s take a moment to unpack the potential in this.
Running AI on a phone means that the entire AI model is stored and processed locally on the device, rather than relying on cloud servers. DeepSeek’s distilled version is lightweight enough to function on a smartphone’s hardware, making it possible to perform complex tasks like text analysis, language translation, and even contract review without needing an internet connection.
This is a stark contrast to apps that connect to cloud-based AI models like ChatGPT. In those cases, the app acts as a front-end interface, sending user inputs to remote servers where the AI processes the data and sends back a response. While this approach works well for users with reliable internet access, it creates barriers for those in areas with poor connectivity or expensive data plans.
This has huge implications for global education, particularly in underserved regions where expensive hardware and subscriptions are out of reach.
Imagine a student in a remote village using their phone to practice English with an AI tutor, analyse complex texts, or even prepare for standardized tests. DeepSeek’s open-source nature and low hardware requirements could democratize access to high-quality education, leveling the playing field for millions of learners worldwide
How Is It Different from Cloud-Based AI Apps?
Offline Accessibility:
I mentioned this before but it is worth repeating! With DeepSeek running locally on a phone, users can access AI-powered tools anytime, anywhere—even without an internet connection. This is a game-changer for students in remote or rural areas, where internet access is often sporadic or prohibitively expensive. Imagine a student on a long bus ride practicing vocabulary or going through a text with an AI tutor, all without needing to connect to the cloud.Cost Efficiency:
Cloud-based AI services like ChatGPT often come with subscription fees or usage-based pricing. By running locally, DeepSeek eliminates these costs, making AI-powered education tools more affordable for individuals and institutions. This could enable schools in low-income regions to integrate AI into their curricula without straining their budgets.Data Privacy and Security:
When AI runs locally, user data stays on the device, reducing the risk of privacy breaches or unauthorised access. This is particularly important in educational settings, where sensitive student information must be protected. Also, this would be invaluable with the data rights of young learners. In contrast, cloud-based AI systems require data to be transmitted to external servers, raising concerns about data security and compliance with privacy regulations.Customisation and Control:
Running AI locally allows developers and educators to customise the model to suit specific needs. For example, a language learning app could fine-tune DeepSeek to focus on regional dialects or industry-specific terminology. This level of customisation is harder to achieve with cloud-based models, which are often one-size-fits-all.Reduced Latency:
Local processing means faster response times, as there’s no need to wait for data to travel to and from remote servers. This creates a smoother, more responsive user experience, which is crucial for interactive learning applications.
The Future of AI in Education?
By removing the barriers of cost, connectivity, and infrastructure, it has the potential to make high-quality learning tools accessible to everyone, everywhere.
But there are still many issues to overcome. Over the weekend I’ve been working with Deepseek, it is clear there is censorship of topics (which I’ve been told are easy to bypass by creating your own fork of Deepseek) and other issues with copyright that may slow the growth its adoption in education. I’m excitied to see how this fasI’m excited to see how this fast, low-cost, and most importantly open-source AI progresses in the future!
