When American-made products become a necessity to global audiences, other countries almost always attempt to replicate its success. It’s only a matter of time before one surpasses the original.
DeepSeek AI, a recently released Chinese generative AI model, has led to mass controversy for the future of technology. I recently joined the market for a more innovative AI model when I discovered that 550 tons of carbon dioxide were produced simply training ChatGPT’s GPT3 model. That is the equivalent of over a million miles driven by gasoline-powered cars. This alarming statistic motivated me to quit my already minimal use of the model until a more efficient choice was created.
DeepSeek AI answered my prayers. It was showcased as an energy efficient, cheaper and smarter alternative to the traditional GPT model. Because the model was made in China, DeepSeek had no access to the American-made, high quality semiconductor chips made by ChatGPT’s parent company Nvidia. This forced the developers to create multiple optimization techniques at a cutting price. DeepSeek’s R1 model cost $6 million to train, contrasting billion dollar investments for OpenAI’s o1d model.
However, the feature that intrigued me the most was its use of the auxiliary loss free strategy. It’s a way of selecting which parts of the model to train instead of training the model in its entirety – which is the standard for ChatGPT. This strategy massively reduces energy consumption and provides a more sustainable and replicable model for an otherwise egregiously wasteful industry for our planet. With this information at hand, it was time to put DeepSeek to the test.
I believe the most effective way of analyzing an AI model’s proficiency is by writing a series of prompts with varying difficulty and ingenuity. I asked prompts regarding basic, immediate and creative context handling. The comparison left me impressed. Not only was DeepSeek more verbose than ChatGPT, I also found it could retain critical contextual information for longer periods without needing to rephrase or ask for clarification.
Ultimately, I do not see an ethical use for AI. AI threatens creativity, encourages propaganda and openly repurposes content already available on the internet uncredited. However, after comparing the two models I choose DeepSeek because I empathize with our posterity.
We are in a dark moment for the planet’s future. This weather is abnormal and unprecedented. The repercussions can be felt when the air conditioning is on during colder months. This January was officially the hottest on record; a severely grim accomplishment. When society begins to shift from harmful energy dependents like ChatGPT, only then will there be a chance to preserve the planet for our future.
