Dear Startups: The power of AI is not in the technology, but in its use to well-defined problems

Gabriel A. Silva
4 min readDec 2, 2023
Image credit: Getty

While my day job in academia is to understand the physiology and algorithms that the brain uses in its computations, some of the work I do is with startups and larger companies. Much of what I do with companies is at the intersection between highly advanced scientific and technical research, and advising and working with companies to help them strategically define their problem and application spaces by leveraging machine learning (ML), artificial intelligence (AI), and other technologies.

In particular, exploring the intersection between startups and the rapidly evolving landscape of ML and AI is critical. The remarkable pace of advancements in this field, exemplified by technologies like ChatGPT and other large language models (LLMs), underscores the importance for startups to meticulously identify and understand their problem space before diving into research, technology development, or employing AI for an ill-defined business objective.

Understanding the Problem Space in the Era of AI

Here, my use of the term ‘problem space’ refers to the specific set of challenges and needs that a startup aims to address. In other words, what exactly is their business and what are they trying to achieve? This is much harder than it sounds. I have personally seen examples of product groups within some of the biggest tech companies have no idea why they where doing what they were doing, or even who their intended customers were supposed to be.

In the context of AI, this is not just a backdrop but a dynamic and integral part of the startup’s strategy. The rapid evolution of AI technologies, such as OpenAI’s ChatGPT, has transformed the business landscape, making it imperative for startups to have a clear grasp of the problems they aim to solve. This clarity helps in aligning their efforts with the capabilities and potential applications of AI technologies. AI tools are not always the answer. Other technologies and mathematical and statistical models sometimes should take priority.

The Importance of Precision in AI-Driven Solutions

But when such tools should be (part of) the strategic solution to a company’s trajectory, ML and AI technologies, while powerful, are not magic wands or oracle-like black boxes. They will not magically solve a problem that is poorly defined or understood. And they will not define the problem for you in the absence of a foundational understanding and appropriate guidance.

Their effectiveness is significantly enhanced when applied to well-defined problems. Startups venturing into AI need to be precise about their target market, the specific pain points of their customers, and how their solution fits into the existing ecosystem. A misalignment here can lead to ineffective solutions or, worse, the amplification of existing problems.

A look at recent startups in the AI field reveals a pattern. Success stories often come from companies that had a clear understanding of their problem space and how AI could be leveraged to address it. Conversely, startups that struggled often lacked this clarity, leading to misdirected efforts and resources.

The Role of Research and Development

In the dynamic landscape of AI-driven startups, the role of research and development is pivotal, extending far beyond traditional product development to form the foundation of aligning technological capabilities with specific industry challenges. Research and development in startups must focus on the contextual application of AI, tailoring technologies like LLMs or neural networks to address unique industry challenges. Critically, this almost always necessitates domain expertise and unique experiences specific to their target business and intended markets.

This could involve conducting early feasibility studies to assess AI’s applicability in identified problem spaces and embracing an iterative approach where prototypes are continuously developed, tested, and refined. Effective research and development teams necessarily comprise a mix of AI experts and domain specialists, ensuring a user-centric design by collaborating with end-users, industry experts and, when appropriate, leveraging academic collaborations and experts. Companies must stay on top of the latest advancements in AI and ML, integrating new findings into their work while focusing on scalability and future-proofing solutions.

Ethical considerations are integral to this process, ensuring that AI solutions are fair, transparent, and respectful of privacy and data security, with a focus on sustainability to assess the long-term impact of these solutions. This multifaceted approach to research and development is critical for developing AI solutions that are not only innovative but also practical, ethical, and capable of making a meaningful impact in the intended domain, not just adding to the fluff and noise.

The Pitfalls of Ill-Defined Technology-Led Approaches — And How to Avoid Them

A technology-led approach, where the focus is on the capabilities of AI without a clear understanding of the problem space, can be a critical pitfall for startups. This approach can lead to solutions in search of a problem, rather than solutions addressing a well-defined need.

For startups in the AI field, building a team that understands both technology and the problem space is crucial. This team should include not just technologists but also domain experts who can provide insights into the specific challenges and needs of the target market.

As AI continues to advance, the importance of identifying and understanding the problem space will only grow. Future technologies will offer even more powerful tools for solving problems, but their effectiveness will be contingent on the precision with which those problems are defined. (Just wait until fault tolerant quantum computers come online.)

For startups in the rapidly evolving field of AI, understanding the problem space is more critical than ever. As ML and AI technologies become more sophisticated, their potential impact grows. However, this impact is contingent on the ability of startups to clearly define the problems they aim to solve. The success stories of tomorrow will belong to those who understand that the power of AI is not in the technology itself, but in its application to well-defined and carefully considered problem spaces.



Gabriel A. Silva

Professor of Bioengineering and Neurosciences, University of California San Diego