Generative artificial intelligence stands at a crossroads: it can evolve from being a mere feature into a full-fledged product with significant economic value. However, current technological limitations and high costs remain considerable barriers.
It’s hard to disagree with Stephanie Kirmer when she states that generative AI (GenAI) is at a turning point (The Economics of Generative AI, in Towards Data Science, August 1, 2024). The key question is whether we deal with a feature or a product. If GenAI is a product—or can become one within a reasonable time frame—it holds enough intrinsic value for people to find it worth purchasing. This is no trivial matter. Having advanced technology is one thing but transforming it into a product that people will buy—and creating a sustainable and renewable business model from it—is another.
Take something as well-known as ChatGPT. OpenAI aims to market it as a mass-market product. But is it worth spending $20 a month to access it? If the answer is no, then ChatGPT is in trouble. The risks are far lower if GenAI is not a product but an additional selling point. Consider Google Search and its, so far, underwhelming attempt to add AI-generated summaries to the traditional list of relevance-ranked results. If this feature isn’t popular, it can be “turned off,” allowing users to revert to the old search model. Apple has embraced the philosophy of integrating existing and future product lines to make the iPhone more beneficial than selling a new model as a standalone product.
A similar discussion applies when moving from the consumer market to enterprise applications, where Spindox operates. The question remains: is GenAI a feature or a product? Today, the answer is straightforward: except in rare cases, GenAI does not currently have enough intrinsic value to justify its commercialisation (and purchase) as a product. That may change tomorrow. Much will depend on training new models to improve their performance. But training has a cost that few can afford—at least until we find a way to do it with a reasonable amount of computing power. A cost that may even be excessive for big tech companies. OpenAI could lose up to $5 billion this year, according to an analysis by The Information based on undisclosed financial data and insider sources (Amir Efrati, Aaron Holmes, Why OpenAI Could Lose $5 Billion This Year, July 24, 2024). If this figure is accurate, the company—valued recently at $80 billion—will need to raise additional funds within the next 12 months.
According to David Cahn (partner at Sequoia), “AI CapEx” is a euphemism for building physical data centres involving land, energy, steel, and industrial capacity. We are now in a cycle of competitive escalation among three of the largest companies in history, collectively valued at over $7 trillion (The Game Theory of AI CapEx, July 16, 2024). Cahn is not concerned about this. However, according to Goldman Sachs strategists, investors are increasingly worried that U.S. tech hyperscales (Amazon.com Inc., Meta Platforms Inc., Microsoft Corp., and Alphabet Inc.) are overspending on artificial intelligence. Over the past year, these mega-cap companies collectively spent around $357 billion on CapEx and R&D. “Given the focus and architecture of GenAI technology today […] truly transformative changes will not happen quickly, and few—if any—are likely to occur within the next 10 years,” said MIT’s Daron Acemoglu in Goldman Sachs’ Top of Mind report (GenAI: Too Much Spend, Too Little Benefit? June 27, 2024).
This is what we explain to our clients. Today, we adopt a perspective that incorporates GenAI within a broader framework. The solution we design should align with the client’s business objectives and encompass several components, including domain expertise, the ability to model the problem, and a comprehensive understanding of artificial intelligence. This understanding includes predictive inference through machine learning, causal inference, mathematical optimisation, and new content generation. Additionally, it requires reliable and secure data management, integration with the wider context of information systems, and enhancing user experience.
All these elements work together within our decision intelligence platform, Ublique.ai, which orchestrates data flow and integrates all necessary analytical models and algorithms. Within this framework, GenAI can effectively deliver value.
