DeepSeek: Is This a “Sputnik Moment”?

When the Soviets launched the Sputnik satellite in 1957, it marked a leap forward in space technology, triggering a “space race” with the American space program.

Source: Alexandra_Koch/Pixabay

In January, Chinese company DeepSeek debuted DeepSeek-R1, its powerful AI large language model. Following its release, tech stocks tumbled as share values declined sharply for chip makers, software firms, and even energy companies. In the quickly evolving field of artificial intelligence, the question is: “Are we facing another Sputnik moment?” What is happening in the AI technology race, and why did a new digital assistant, among the dozens that are already available, shake up the global computing industry?

Artificial intelligence is the field of building computers that simulate human intelligence. There are numerous AI models available: algorithms trained through machine learning from large data sets to recognize patterns and make decisions without human input.

DeepSeek offers several key innovations. First, this AI model was shared as “open-source,” meaning the model was made publicly available for others to use. Next, and most importantly, this AI model was purported to accomplish many of the same tasks as other competing models but with much lower CPU computer requirements and energy demands. DeepSeek is a new AI model that can use less specialized chips and less massive computing and energy infrastructure. Due to this efficiency, the upstart, open-source DeepSeek AI model represents a threat to existing information technology supply and demand.

This innovation took investors by surprise and triggered a cascading impact on global semiconductor, software, and energy stocks. Given the ensuing media excitement and even calls to ban or regulate the new model, it now appears that we have entered an age where each new AI discovery will fuel the AI “space race,” producing many future “Sputnik moments” and potential market disruptions.

One use of the new supercharged AI assistants is to ask them to sift public information and summarize news and information. For example, an AI bot could write a blog post such as this one. This particular article was written by a human author, but you wouldn't necessarily be able to tell that just from reading it. That is because AI is getting very good at generating content that is virtually (no pun intended) indistinguishable from material created by humans. AI bots perform research and answer “prompts” or questions about any topic by accessing the vast wealth of knowledge on the internet and the World Wide Web. “Generative” AI is even more creative, learning from videos and artwork to generate realistic graphics, photos, or even deep fakes, limited only by a user’s imagination.

Research and creative writing are just two examples of how to use these AI models. How about conversation? Most of us have already interacted with an AI customer service chatbot. These programs are becoming ever more sophisticated, helpful, and realistically conversational. But human-AI interactions do not stop with customer service questions. Increasing numbers of adults now rely on daily conversations with their AI chat buddies. As we share personal information with a chatbot, the LLM learns more about us, and the AI-human interactions begin to mimic genuine human relationships. Users who are comfortable with technology say they are open to viewing their AI assistants as “friends,” and some even acknowledge that AI friends have replaced some personal—and even romantic—relationships. (1,2)

Many remain skeptical of the close bonds humans can forge with digital AI counterparts as these entities become more common and “human-like.” Our views about relationships in a digitized world may have to evolve. AI companions can be programmed to be nonjudgmental and supportive and tell us what we like to hear. Some of us are concerned that users could become socially isolated, avoiding human interactions. However, AI companions might be useful for those with mental health issues, helping to offset loneliness and providing a safe space to develop needed social skills. AI models are already surpassing simple applications and are evolving to the point where we can develop emotional bonds with our AI companions. (1,2)

AI models serve a variety of purposes: to converse with a chatbot, create music or voice from text, track numbers, create and edit images, generate content, program software, save time, and improve productivity. One of the most compelling things about AI is that it may eventually be able to teach itself new tricks. In other words, these models can continue to learn and improve. In the future, AI models will enhance their own intelligence to the point where these models achieve super artificial general intelligence (AGI) equal to or even surpassing that of humans. No AI has achieved AGI yet, but this level of intelligent machine is on the horizon and likely to emerge soon.

Super-smart AI may be used to discover new medical treatments, advance scientific discoveries, and provide the opportunity to communicate with a new intelligence. Yet, as shown by the recent stock market tumble following the release of just one new AI product, we are entering an age of both risk and reward. AI is evolving rapidly and can impact society at all levels. This technology's potential to replace jobs is just one concern. What happens if AI replaces friends and partners?

Are we ready for that?