{"id":10177,"date":"2025-02-13T09:00:00","date_gmt":"2025-02-13T09:00:00","guid":{"rendered":"https:\/\/blogs.sussex.ac.uk\/tel\/?p=10177"},"modified":"2025-02-26T15:26:46","modified_gmt":"2025-02-26T15:26:46","slug":"deepseek-ai-advancement-and-the-path-forward-for-ai-in-education","status":"publish","type":"post","link":"https:\/\/blogs.sussex.ac.uk\/tel\/2025\/02\/13\/deepseek-ai-advancement-and-the-path-forward-for-ai-in-education\/","title":{"rendered":"DeepSeek, AI advancement and the path forward for AI in education"},"content":{"rendered":"\n

The world of AI is a fast-moving place, shown very clearly by the arrival of a new AI-model from a Chinese company called Deepseek a few weeks ago. The whale in the company’s logo is very fitting as the release of their V3 model certainly made a splash!\u00a0\u00a0<\/p>\n\n\n\n

Although perhaps a shark logo would have been more appropriate given the impact over the last few weeks. The release of the V3 model immediately wiped almost $600bn (\u00a3482bn) off the market value from Nvidia, the biggest one-day loss in US history, and set panic amongst American AI companies and AI policy. But why the fuss and how does this relate to the future of AI in relation to education?\u00a0<\/p>\n\n\n\n

\"\"<\/a><\/figure>\n\n\n\n

Why is it such a big deal? <\/h2>\n\n\n\n

There were a few reasons this release was such a big deal, including its low cost, its power-usage and the fact it\u2019s been made open-source. Let\u2019s explore what each of these points mean.\u00a0<\/p>\n\n\n\n

Cost <\/h2>\n\n\n\n

DeepSeek was reportedly trained on a budget of 6 million dollars, a fraction of the cost compared to the American models. DeepSeek V3 is comparable to the most powerful models that competitors like ChatGPT have developed and, more importantly, DeepSeek is able to give access to their model for free. You would have to pay almost $200 a month to get access to ChatGPT\u2019s comparable model.\u00a0<\/p>\n\n\n\n

From testing, the results these are essentially the same. OpenAI has a few more features but the models processing and speed are roughly on par.\u00a0<\/p>\n\n\n\n

Power usage <\/h2>\n\n\n\n

AI models need a lot of processing power to train them. The best way is to use GPU\u2019s (Graphic processing units) which are able to do advanced processing of data. Now America has restricted Chinese access to Nvidia\u2019s H100 GPU\u2019s because of their usefulness to training AI models. By restricting them, the American policy aimed to prevent China developing strong AI models…or so it was thought!\u00a0<\/p>\n\n\n\n

But necessity is the mother of invention and with Chinese companies like DeepSeek being forced to use much weaker GPU\u2019s (they used Nvidia\u2019s much slower H800 GPU\u2019s), they were able to work out ways of working smarter, programming models that got the same result using much less processing power. DeepSeek researchers then published a paper<\/a> showing how they achieved this result and how others can do the same.\u00a0<\/p>\n\n\n\n

Open source <\/h2>\n\n\n\n

DeepSeek\u2019s model has also been released as open-source. The model can be used for free by anyone locally which means it would run entirely from an individual’s own computer (there is a cut down model that can be hosted on a reasonably powerful computer and a full model which requires a very powerful computer to run. It\u2019s still out of reach for most people but closer than it was). This open-source approach is in opposition to other AI models such as OpenAI Google and Microsoft\u2019s which are closed. This means you cannot run your own version of it but can only run it from those companies’ servers.\u00a0\u00a0<\/p>\n\n\n\n

Running your own model means you have complete access to how your data is used, plus you aren\u2019t beholden to a company which has now clear advantages- we\u2019ll cover that later.\u00a0<\/p>\n\n\n\n

Should we be using it?<\/h2>\n\n\n\n

There\u2019s no real reason to be less trustful of DeepSeek then any other model. DeepSeek was created in China and so has some censorship built into it due to the Chinese government, however American companies’ models have strong connections to the current American government and their own biases and potential data issues.\u00a0\u00a0<\/p>\n\n\n\n

You could argue that DeepSeek is currently the most \u2018trustworthy\u2019 (used loosely) of the big models or at least the best of a bad bunch given their release of an open-source model and free sharing of their research compared to the American companies’ secrecy. Also, at the time of writing, the only one to not have a contract with defence or defence adjacent weapons companies (Google has recently withdrawn their policy against developing weapons systems). DeepSeek also uses the lowest amount of power to run queries making it the most \u201csustainable\u201d (again used loosely) of the current crop of models.\u00a0<\/p>\n\n\n\n

We should be using the same critical thinking when using any AI model regardless of which company it comes from<\/strong>. This means you shouldn’t put any private or personal information into any AI model that\u2019s not been approved for use by the institution. Don\u2019t download any AI tools directly to a computer unless it\u2019s been approved. Assume all models will have biases and potentially be aligned to certain views and values. Don\u2019t assume anything it tells you is automatically true- it\u2019s healthy to apply this principle to any AI model you use regardless of where it\u2019s come from.\u00a0<\/p>\n\n\n\n

The ideal path forward for AI models at Universities <\/h2>\n\n\n\n

DeepSeek does however point us in the direction of a more ethical and safer future for AI- smaller, more efficient open-source models that are locally hosted and run by individuals or institutions, entirely under their own control without the data being taken or used by an outside company. This would also allow for more ethical models with less bias and censorships. Currently, there are cost barriers but a model like DeepSeek can run on a computer that costs a fraction of what ChatGPT and similar take to run.\u00a0<\/p>\n\n\n\n

The advantages of these models would be multifaceted. First, a decoupling from what are largely unethical and unsustainable companies, meaning that students and staff could use AI models without worrying that their data is being harvested. Secondly, models could be tuned that are more free of biases and censorship.\u00a0\u00a0<\/p>\n\n\n\n

Of course, for this to come to fruition we\u2019d need a few more technological leaps to make even smaller models that use much less power, but DeepSeek\u2019s progress is an encouraging step in the right direction and more competition will hopefully breed more innovation.  <\/p>\n\n\n\n

For better or worse we cannot avoid the proliferation of AI into our society and institutions, and so, taking an optimists view, I\u2019d like to hope we can move to a less corporate ownership model and put such tools in the hands of Universities that can use them for the public good. We can but hope!\u00a0\u00a0<\/p>\n

Share this:<\/h3>