Call it snark or pragmatism, but a recent BioBuzz panel, “Leveraging AI Across the BioPharma Value Chain,” brought something rare to the AI conversation: honesty. The panelists didn’t worship at the altar of algorithms. Instead, they asked smart questions. When does AI really add value? When is it just a high-tech answer in search of a problem? And who actually benefits when every analytical tool gets relabeled as AI?
One thing was clear. The term “AI” has become so vague it’s practically meaningless. What a computer scientist calls AI might look completely different to someone in pharma or biotech. And that disconnect creates a perfect breeding ground for hype, confusion, and more than a few overpriced tools. As one panelist put it: “AI is cool, but it’s not good for everything.”
There are different kinds of AI, and they serve different purposes:
- Natural language LLMs, like ChatGPT, are great for content generation and idea synthesis. They’re accessible to non-experts and can be surprisingly helpful for the right tasks.
- Reasoning models overlaid on LLMs add structure and logic, but still require guidance.
- Generative models, like those used for protein design or antibody discovery, enable genuinely new science, but they are not self-driving. They belong in expert hands.
In short, context matters. An LLM might help draft an internal IT policy. But AlphaFold won’t mean much unless you understand structural biology.
Still, in today’s climate, AI gets slapped on nearly everything. Logistic regression? Random forests? Basic statistical models from the seventies? Rebranded and marked up. Why? Because “AI” sells better.
At Diamond Age, we see this play out regularly. Clients come to us looking for an AI-powered solution, but the dataset isn’t strong enough to support it. In those cases, the most responsible thing we can do is to be honest. We’ve had to tell clients, after exhausting every option, that no amount of modeling or machine learning will save a weak dataset. The most helpful path forward is often to generate better data—not throw more algorithms at the problem.
That’s the piece too often forgotten. AI is not magic. AI does not innovate. People do. Scientists with domain expertise turn tools into insights. AI can suggest possibilities, but it can’t tell you which ones matter. That requires human judgement.
We use AI when it makes sense. But often, simpler tools solve the problem faster, better, and for a lot less money. Before reaching for the flashiest model, we ask if there’s a more direct route.
And here’s what makes us different. We’re not selling AI. We’re selling expertise. That means we’ll always tell you the truth, even when it’s not what you hoped to hear. Sometimes the most valuable thing a consulting team can say is, “You don’t need AI for that.”
It’s not just good science. It’s good business.