DeepSeek | China's Artificial Intelligence that is challenging Mayor Players such as OpenAI

Technology has evolved rapidly around artificial intelligence. In fact, if there is anything to mention, it is that this is the key factor of this era, so surely in the following years, all technology we know today will be adapted to AI. This makes us assume that there will be new advances that will also evolve by leaps and bounds, which from a perspective is a good thing, because we will be able to experience that quantitative leap in just a few years. However, we do not take into account that this growth is becoming unrestrained. For if you look around, AI is covering everything; even in what you are good at.

Some great technology experts told us that we could be replaced by machines. Indeed, and to some extent it is happening. Robotics comes along with the potential of AI, and if the latter achieves an extremely large boost, so will the hardware that incorporates this type of models. But this is not the topic we will discuss today, but rather the potential to excel beyond the rest of the intelligences of a new reference.

Lately, a new AI trend is emerging, with the difference that it does not come from American companies, but from the Asian country of China. DeepSeek is an artificial intelligence that is making waves by managing to compete against the leading companies that most of us know today. It stands out more because it uses much smaller models, but with great efficiency when using data. This suggests that the resources needed are less than those consumed by companies like OpenAI. Also, it's incredible that a company founded in 2023 is already on par with its most potential competitors; for example, OpenAI and Anthropic.

It is really interesting, as I mentioned before, that this new company has an AI processing power at the level of GPT-4. Although its way of solving requests is very different, it effectively balances the way it manipulates data to achieve exponential performance. By the latter, I mean that this AI is guided by inference (model R1), which is nothing more than consulting the data with which the model has been trained, to continue learning on the go.

The commotion that occurred in the last few hours has been very noticeable, especially on the well-known and popular social networks of today. The mobile applications of DeepSeek have managed to be in the top position in the ranking, both in the Apple Store and in the PlayStore for Android. In fact, if you take a look at this precise moment, you will confirm what I am saying. It has already surpassed 10 million downloads, with a rating close to 5 points.

This might seem little, but we must consider that this is not an application from the West, but rather a Chinese competitor that is starting to excel in many of the most important aspects, such as costs, performance, and efficiency. Taking into account that there are certain restrictions when it comes to acquiring chips, we can point out that with technology that is not cutting-edge, or in some respects worse, they have achieved something unimaginable.

From my point of view, if the market were in favor and allowed access to cutting-edge AI chips so that the models could be supported, they could advance in an unrestrained manner. But this means that the leadership of companies like OpenAI will be forgotten, because a new technology is taking center stage. At this point, there is a disagreement that is heightening discussions, or at least generating certain tensions. DeepSeek is accused of having used the OpenAI API without permission. In fact, in the company's terms of service, it is stated that the API cannot be used for training and creating new AI models; this aspect is categorically denied. This is something that Microsoft had to point out to OpenAI after investigating the case, as it is one of the largest shareholders of this company.

Returning to the issue of DeepSeek's potential, one thing is certain is that it is impressing many, although not everyone will be pleased. Bearing in mind that power and efficiency can be achieved, along with precision, with fewer resources than imagined, is something that caused a drop in the shares of a particular company. NVIDIA, a company that manufactures cutting-edge, but high-cost chips, lost over $600 billion dollars after DeepSeek released its R1 model, reflecting that raw power is not required to do incredible things and compete with its rivals. The outlook is not very encouraging for the West unless they promptly evaluate how not to fall behind in this technological competition.

• Main image edited in Canva.
• I have consulted information at techcrunch.com : I, II, III and IV.
• Translated from Spanish to English with Hive Translator.

Posted Using INLEO