DeepSeek is a new language model with code released under the MIT License. To keep things simple, what this means is that DeepSeek is much more an "Open AI model" than any model produced by OpenAI in recent history. There is something to be said about the evolution of OpenAI from more ethical beginnings into its current form, and I may write more about that in a follow-up to this post.
There are multiple notable achievements presented by DeepSeek. It performs at a level that is roughly equivalent to or one that surpasses most similar properitary models, including those of OpenAI and Google. It was developed as a "side-project" for a budget much smaller than those of the corporate entities mentioned above; similarly, it uses much less computing power. But perhaps most importantly, it was developed and released as an open-source project, with the stated goal of creating long-term social value.
A single open-source AI model exerting such an influence so as to crash nearly every AI-related stock in the U.S. market is a statement about value. To me, this is an embodiment of enshittification and an enormous victory for both the free software and open source movements. It is clear that the companies who have financially suffered as a result of this abandoned the capitalist pillar of innovation - the creation of outstanding value - long, long ago.
A large issue pertaining to the value of proprietary software is that the primary motivation will always be revenue. Most of the time, if there is not a justifiable projected increase in revenue, beneficial updates will not be pushed to users. In fact, much of the time, useful features are removed - updates which make the product worse are pushed. Think of this: it is someone's career, someone's profession at an organization to actively make the product worse for customers. This is true not only of the tech giants, but also of the companies who proclaim themselves as pioneers, beacons of light, and revolutionaries of the free software / open source movements. Proton is an excellent example of a company which puts on a facade but really pushes updates that consistently enshittify their systems.
There is typically a tradeoff between value and convenience, if not the illusion of convenience that familiarity grants, when one is choosing between proprietary and non-proprietary products. For instance, if one is choosing between Windows and Linux, there are clear winners in most cases. A non-technical user will almost always choose Windows for the familiarity that it provides, without considering the alternative. Many of the most technical users become cultlike with their Linux distributions of choice because of the perceived value that it provides. In terms of which system treats the user better and consistently introduces more value, Linux is a definitive winner. Windows maintains its dominance not through value, but by through familiarity and brand-recognition.
Why, then, is a relatively unheard of model produced by a definitely unheard of company able to cause such a disruption? It is because the true innovation and value provided are magnitudes ahead of what even the properitary competitors can provide. DeepSeek can be deployed locally, free-of-charge, on consumer-grade hardware. If one were to consider the performance per capita given the resources required to run the model, DeepSeek V3 is an unprecedented innovation.
Though it may sound extreme, I argue that this provides a strong example of the ethical shortcomings of the capitalism system as implemented in the United States. If it is to be that the expectation is that all members of a society contribute to it, then we must consider corporations, considered by the law to be a consolidated entity, as a member. Thus, corporations must be compelled to contirbute to the overall benefit of society.
This goes to show that the markets do not respond to innovation. It further demonstrates that the ideals of corporations are not those of contribution towards principles of mutual benefit. Rather, the only clear motivation is that of profit. This could be seen as a form of utmost greed - the abandonment of innovation, of principles that are instilled in us as utterly American, for a quick (and clearly unreliable, in light of these events) dollar.
I sincererly hope that this will bring some renewed interest both academically and by the general public to the free software movement. Though the MIT License is not a copyleft license, DeepSeek is nonetheless a remarkable contribution to the free world. It is my hope that rather than resulting in the public being immersed in media hysteria over the events, it is an eye-opener.
The battle against technological injustices, against corporations who prioritize greed and profits and forsake innovation and contribution to the progression and wellbeing of society, is not a new one. This moment in time can serve as an important introduction to the ideals that we are fighting for. It can serve as a living example of the outcomes which would be possible if only more were to join our cause, join our rejection of unethical ideals in favor of moral ones.
Many people, the open-source folks included, see the free software movement as too idealistic, as impossible to implement. I argue that this event is a major rebuttal to said arguments. Onward, then!