Stop Chasing Technology: Finding the user’s problem is the real problem!

There’s a common critique in the AI community: too often, AI solutions emerge
from asking, “What can we do?” rather than, “What problem do we need to solve?”
Developers frequently jump on the latest algorithm or innovative model and then
search for a suitable use case. This approach tends to produce superficial
fixes—flashy AI wrappers that look impressive on paper but fail to address the
complex, real-world challenges faced by businesses and consumers.

I firmly believe that innovation should follow a different path: start by
identifying a genuine problem and then apply even a suboptimal solution to
address it. Discovering the right problem is far more critical than finding the
perfect solution from the outset. If a solution truly tackles a real issue, it
creates the foundation upon which more advanced and refined approaches will
eventually build.

Consider the example of the Wright brothers. Their early flying machines were
simple, even rudimentary, yet they marked the beginning of aerodynamics and
aviation. The modest success of these initial designs paved the way for decades
of innovation, transforming what was once a toy into the sophisticated aircraft
we know today.

The Pitfalls of a Technology-First Mindset

Building AI solutions solely because a new technology is available might seem
appealing at first. However, when developers start with a pre-made solution or
an over-fitted algorithm, they risk missing the nuances of the actual problems.
Without fully grasping the intricacies and constraints involved, any resulting
AI model might scratch only the surface and ultimately produce a product that
doesn’t truly solve anything.

This focus on technology can also divert attention from the real needs of users
and stakeholders. When the emphasis is on showcasing cutting-edge algorithms or
leveraging popular frameworks, the solution might end up as little more than a
shiny AI wrapper—a superficial interface simply built around existing
components.

An example of this can be seen in the current trend where large
language model (LLM) solutions are built around NVIDIA’s ecosystem. While
NVIDIA’s hardware and software are undeniably advanced, merely wrapping AI
functionalities around these infrastructures does not necessarily foster true
innovation. Instead, it reminds us of the dotcom boom, where companies
rushed to build an online presence without a clear business model, or the
crypto hype, where flashy technology promised fortunes but often failed to
resolve any substantive issues.

The tech world has seen AI wrappers fall short by reusing old ideas in new
covers. Chatbots often fail as robotic because they don’t integrate well
with other systems. AI marketing tools simply repackaged existing data without
adding much value. Similarly, hiring systems that were meant to improve resume
screening ended up missing subtle human skills, leading to overlooked talent.
Their reliance on fixed rules made them fragile, much like the tech that failed
during the dotcom crash.

Problem-First AI Development is the way to go

For AI to truly matter, focus on the problem first. Instead of adding AI to
flashy but shallow systems, start by understanding the challenge and defining
its key issues. This approach builds tools that really meet user needs. Engage
with end users to learn their pain points and collect high-quality data for
robust models—much like the sustainable methods that helped companies recover
after the dotcom crash. Finally, consider ethical and social impacts; this
careful evaluation builds trust and ensures your AI evolves responsibly and
benefits everyone.

Conclusion

Generative AI expands our possibilities, but we must focus on real needs.
Instead of chasing flashy tech, we should tackle clear challenges, listen to
users, use quality data, and care about ethics. As the dotcom and crypto booms
showed, true success comes from solving real problems, not just seeking for a
problem to apply new technologies and algorithms.