DeepSeek: The Geospatial AI Genie is Out of the Bottle

The emergence of DeepSeek, an open-source generative AI solution from China, is a seismic event reshaping global perceptions of AI dominance, investment strategies, and the accessibility of cutting-edge technology.

DeepSeek also has huge implications for Geospatial AI.

What We Have Learned So Far

So why has this DeepSeek announcement sent such ripples through the AI industry?

Here is what we know so far:

• DeepSeek was developed using the less powerful Nvidia H800 chip (the US has banned China from access to the newest most H100 Nvidia chips)
• The project cost $6m. OpenAI's CEO, Sam Altman, has stated that training GPT-4 cost more than $100 million. While Dario Amodei, the CEO of Anthropic, put the current cost of training advanced models at between $100m and $1bn.
• The DeepSeek code has been open-sourced. So, download and run locally.
• The DeepSeek AI Assistant app, powered by the DeepSeek-V3 model, can be downloaded for free from both the Apple App Store and Google Play Store. So a sophisticated AI model which runs on smartphones.

Implications

DeepSeek challenges the notion that advanced AI requires massive financial and computational resources. Liang Wenfeng, the founder of DeepSeek, has embraced a vision where AI is free and accessible to all. This has triggered doubts about the sustainability of AI’s current economic model. Currently in the US, AI is largely proprietary (Meta's Llama 3 is an exception), development being driven by the powerhouses of the Web 2 world (Google, Microsoft etc). DeepSeek disrupts traditional profit-driven models, decentralizing innovation and empowering grassroots solutions to local and global problems.

What this implies is the democratization of AI tools, putting sophisticated capabilities into the hands of everyday users, enabling innovation beyond the control of tech giants.

That makes DeepSeek potentially a game-changer.

Ethical and Regulatory Challenges

There remains plenty of ongoing discussion in the area of ethics and regulation. DeepSeek simply adds to this conversation. With such powerful AI technology freely available, misuse, misinformation, and lack of oversight are valid concerns. Unlike proprietary models, where companies can implement safeguards, open-source tools like DeepSeek put responsibility directly on the users.

On the flip side DeepSeek is commercial bias free. It takes 'big brother' out of the equation.

Implications for Geospatial 2.0

AI is at the heart of Geospatial 2.0. It is the backbone and glue of all geospatial innovation as we move forward. DeepSeek potentially has a huge part to play in that evolution:

• Enhanced Accessibility for Geospatial Analytics: DeepSeek’s advanced reasoning capabilities could unlock geospatial insights for smaller companies and individuals who previously couldn’t afford costly AI solutions. This democratization may lead to a surge in innovative geospatial applications.
• Faster Data Processing and Integration: The open-source nature allows for customized applications, especially in processing complex geospatial datasets for tasks like climate modeling, infrastructure planning, or urban development. Open-source AI could significantly lower barriers for real-time, large-scale geospatial data analysis.
• Catalyzing Decentralized Geospatial Innovation: With tools like DeepSeek in play, geospatial data could see broader, decentralized innovation. This supports the Geospatial 2.0 vision, where technologies like AI, IoT, and digital twins converge to create actionable, scalable solutions.
• Challenges to Proprietary Geospatial Platforms: Much like OpenAI faces competition, traditional geospatial leaders like Esri could see market disruption. Open-source AI will likely enable new entrants to the geospatial sector, bringing competition to previously monopolized areas of mapping, spatial intelligence, and data visualization.

Final Thoughts:

In my 25 years of working with technology, I have never witnessed a period of transformation as rapid as what we’re experiencing today. It’s hard to believe ChatGPT launched just two years ago, given how far we’ve come in such a short time. DeepSeek is yet another remarkable chapter in this unfolding AI revolution. The repercussions will continue for some time.

In the context of Geospatial 2.0, DeepSeek aligns perfectly with the paradigm of decentralization and convergence, promising to accelerate innovation. Whether it’s analyzing satellite imagery, processing 4D geospatial models, or automating decision-making processes, tools like DeepSeek might prove to be pivotal in reshaping the future of geospatial technology.

Matt Sheehan is focused on geospatial advances. He writes and speaks about the geospatial paradigm shift – Geospatial 2.0. Matt approaches technology with a strategic business driven mindset. You can reach Matt at [email protected].