Their assumptions are understandable. After all, global tech giants and governments are putting billions of dollars into artificial intelligence. Microsoft recently announced a $300 million investment in South Africa’s AI infrastructure, IBM created a $500 million enterprise AI Venture fund, while China’s honor promises $10 billion to develop AI-driven devices over the next five years.
Even governments are actively investing, and the EU has allocated billions of dollars to AI research and development through programs such as Horizon Europe. Given such high-profile spending, it is no surprise that AI is regarded as an elite technology. But does this narrative obscure the reality?
To challenge this view, I do not rely on arguments – I rely on evidence. I started with a laptop and showed executives how I could build a completely autonomous AI model and chatbot in a few hours.
The model I used only has 8 billion parameters, which is part of the size of state-of-the-art LLM (large language model) like GPT-4, but it performed very well. More importantly, I can start customizing it right away – feeding domain-specific knowledge at zero cost and cloud-free dependencies to develop AI agents that suit my needs. I have complete control over my data and the insights and documentation I trained.
The executives were surprised. So I pulled out my Raspberry Pi, a $100 single-board “computer” with all the key components of the motherboard – a simple 2.4 GHz quad-core 64-bit 64-bit ARM Cortex-A76 CPU and Video VII VII GPU – and showed them how they can run on it.
This setting is not flucom. Thanks to innovations from companies such as DeepSeek, open source AI models have become increasingly powerful and easier to access. Their recent releases are accessible and accessible when introducing the DeepSeek-V3 and DeepSeek-V3 versions, suggesting that high-quality AI models can be developed with relatively low computing resources, lower energy consumption and moderate budgets. Its assumption is that only industry Titans can afford AI.
Also Read: How AI Changes the IT Services Industry in India
The rise of open source AI
The rapid development of open source AI is democratizing access to machine learning. In recent years, many generated AI tools have shown that smaller, fine-tuned models can go beyond proprietary alternatives in a specific task.
Open source AI models are driving innovation in the industry from healthcare to financing. For example, Tensorflow and Pytorch have been widely used in medical imaging for tasks such as tumor detection, improving diagnosis speed and accuracy. OpenChem helps researchers develop predictive models for drug discovery.
In the financial field, open source AI models including QuantConnect and Tazama have been adopted for algorithmic trading, risk assessment and fraud detection. These applications allow financial institutions to process large amounts of data efficiently, resulting in smarter decision-making and improved security measures.
Meanwhile, more and more businesses are realizing that they don’t need Openai or Google’s resources to leverage AI, and they can deploy tailored, efficient models at a fraction of the cost. The impact of this transformation is profound. Once the field of multi-billion dollar R&D labs, AI is now available for startups, researchers and individuals. Companies that embrace this reality will gain a competitive advantage – not by surpassing their competitors, but by innovating smarter and faster.
However, there is a warning: developers and users of proprietary or open source AI models should be aware of potential biases in AI output and ensure they comply with relevant regulations. To retain consumers’ trust, they should also maintain data privacy and transparency in AI applications.
Even for today’s “digital transformation” organizations, adapting to technological disruption requires a level of “restructure agility” that often exceeds their existing capabilities. However, it is crucial to be able to quickly detach from temporary dominant but false narratives during times of high volatility, uncertainty, complexity and ambiguity. Find facts, stick to evidence; and begin reexamining the assumptions and frameworks that we believe in and shape our attitudes, such as innovations in technology-supporting technology.
The Internet was a pioneering technological disruption in the mid-1990s. AI is doing the same thing today with faster and deeper consequences.
A new AI mindset
After the demo, I asked the executives a different question: If AI is so accessible and affordable, what should you do today?
The conversation changed drastically. Don’t worry about getting into deep knowledge, risk, dependency or fear needs, but practical applications for executives to brainstorm – automating internal processes, enhancing customer service, and improving risk assessment and decision-making.
The challenge is no longer financial or technical; it is about imagination and execution. This is a lesson we need to learn. AI is not only suitable for Silicon Valley giants. It is built, perfected and applied by people with a moderate budget and good idea.
Now is the consequence of the future we choose to build. So for your tools, readers. AI is your exploration.
[This article is republished courtesy of INSEAD Knowledge, the portal to the latest business insights and views of The Business School of the World. Copyright INSEAD 2024]
