Finding a trustworthy AI consultant
When you’re looking at implementing AI in your mission-driven organization, you might look into hiring an external consultant to help identify use cases, train staff, design a safer implementation, select tools, or plan your roll out. But the internet is swarmed by people styling themselves as AI experts. How do you find a good one?
Start by Asking Whether You Need a Consultant
Your organization may need a consultant if you're developing a comprehensive AI strategy, navigating complex legal or regulatory requirements, or designing custom implementations that require technical expertise your team doesn't already have.
You might need a vendor instead if you've identified specific AI tools or platforms that could benefit your work but require evaluation and implementation support.
If your needs are primarily educational, consider bringing in a trainer instead
And external expertise isn't always the answer. If you're in the early stages of exploring AI possibilities, internal capacity building might better serve your mission. You could join or found a chapter of a professional organization, get a certification (especially an in-person one that targets organizations like yours), or create a user group within your professional network. These options build lasting knowledge inside your team and will make it easier to work with consultants, trainers, and vendors when you get to that stage.
Navigating a Crowded Field
If you've already decided that you need a consultant and looked around, you have likely found that there are many people calling themselves AI consultants and experts. You want to avoid scammers and substance-free gurus, but it's difficult to evaluate experts if you don't already have expertise!
I recommend paying attention to several things when evaluating a potential consultant: their specific expertise (and how clearly they communicate it), whether and how they talk about ethical issues, how they talk about your organization's context, their relevant experience, their mindset, and their online footprint.
Humility and Expertise
Look for someone with relevant credentials and experience who is very clear about what they don't know. For example, I am plenty qualified to write about and consult on AI in mission-driven organizations: I have a terminal degree and a decade of experience. But there is still plenty I don't know, including types of implementations I've never participated in, regulatory contexts I've never worked in, and technical pieces I am not the right expert for. I will tell you so.
If you are talking to someone who never ventures to the edge of their expertise, it is possible that you have met someone more competent than anyone I've ever met, including the top researchers in information science, human-computer interaction, values-centered technology design, and computer science. It seems more likely to me, though, that they are a smidge overconfident :)
I also suggest watching for a few warning signs:
Confident predictions about the future. If someone tells you "AI will kill [job function]!" or is full of polemics like "AI is all hype," you should be suspicious of their knowledge of the technology’s capabilities and their ability to develop a balanced, tailored strategy for your organization.
Lots of "always" and "never." A consultant with many rigid rules may struggle to adapt their thinking to your particular situation.
"AI" used only to mean large language models or generative AI. A consultant who uses the terms interchangeably may be a newcomer to the field who hasn't been paying attention to AI's broader landscape for very long, including computer vision, robotics, recommendation systems, and many other branches that might matter for your work.
Jargon without explanation. If a consultant can't explain a model or its limitations to you without relying on technical language, that's worth noting. Clear communication is a core consulting skill and will be critical for your leadership to develop judgment that will last after the consultant turns in their final deliverable: you don’t want to be paying for more advice every time a new model comes out!
Big promises. "10x ROI in 6 months!" or "Fully autonomous operations by next fiscal year!" indicate a lack of realism and rigor.
Acknowledging Ethical Issues
AI can do cool stuff. AI consulting attracts people who are very excited about the cool stuff. But this excitement can lead to an "I have a hammer and now everything looks like a nail" problem, resulting in implementations that don't consider ethics and risks to the organization's reputation. Getting very excited about what AI can do can distract from understanding what it is in fact good at, what it is not good at, and when using it can even be risky.
My advice: look for someone who is frankly at least a little scared of AI.
Find someone who brings up risks and ethical concerns without you having to ask. Ask a prospective consultant what they think about bias, sustainability, or other values that are close to your mission, and let them talk. You don't need someone who agrees with you 100%, but you are looking for someone who has clearly thought about the issues and has some curiosity about how they might be relevant to your organization.
Curiosity About Context
Your organization's context is central to a good AI strategy. It's critical to find a consultant who understands that and who is not selling a one-size-fits-all solution. A good consultant will ask about your organization's mission, regulatory context, data maturity, funding model, and key processes, rather than jumping right to implementation.
Experience in Similar Contexts
AI performs unevenly across different tasks and settings, a phenomenon Ethan Mollick named the "jagged frontier," which is in part influenced by your organizational context. A consultant with a lot of experience in investment banking is often not the right fit for your environmental nonprofit. Look for someone who has worked in or consulted with at least a few mission-driven organizations, ideally in your sector: nonprofit, government, agencies, or social benefit organizations. The consultant gets bonus points if they have demonstrated alignment with your organization's mission and values.
Functional context can also influence this unevenness. You are unlikely to find someone who has worked deeply in your organizational context and is deeply familiar with each functional area, but you do need someone who is familiar with several functions, rather than someone who specializes in AI strategy for only marketing or only HR.
A Teaching Mindset
It can be profitable for technology consultants to come in, set up a system for you, and then wait for your call when something inevitably breaks or changes. You don't want that arrangement. Find someone who will "teach you to fish." Ideally, you'll find someone who offers training or (perhaps better in some cases) will train your in-house trainers: your HR department, learning and development folks, managers, and executives.
It is good practice to work with an expert annually or after a big shift in the organization, industry, or technology to ensure you're on the right track. But by the end of each engagement, you should have some confidence that you can adapt to smaller changes without needing to pay for advice.
Respect for Human Workers
Focusing on the power of AI without recognizing the superpowers of your human staff puts your mission and your organization at risk. If a consultant you are evaluating talks endlessly about the cool things AI can do without mentioning its limitations or the benefits of human workers, proceed with caution.
References and Public Footprint
Past client references and an online presence can help you assess all of the features I've described. They are especially useful for checking whether a consultant is telling you what you want to hear or whether they indeed have the experience, knowledge, and care for values that you need. If they have a blog, scroll back a bit and check for transparency, nuance, the ability to explain things without technical jargon, and values (both moral principles and priorities) that align with yours.
A Word About Independence
Finally, a note of caution about independence. Consultants who get referral fees or commissions from specific platforms or who sell proprietary tools may have conflicting interests. Custom software or implementations can be great, but prioritize transparency and the ability to export or migrate without hefty fees. If you can do so, consider a strategy-only consultant and work with them to find solutions that your in-house team can set up and maintain.
If your organization is thinking about AI and could use some help figuring out where to start, I'd be happy to talk. I love working with mission-driven organizations, and I hate feeling like I’m out over my skis: I promise to tell you if I'm not the right fit, and I'll do my best to point you toward someone who is! Reach out to karen@drkarenboyd.com :)
—
LLM disclosure: First, I requested a blog post draft this referring to the book manuscript. The voice was whack: overconfident, salesy, and stuffed with a truly comic amount of em dashes.
Then I tried again with Opus 4.6 by pasting in just the relevant section and asking “can you lightly rewrite this as a blog post?” Still not the right approach, but a bit better. Then I responded in that thread: “can you bring the voice closer to one in the book while keeping therest of it? remove all emdashes, hypophora, rhetorical contradictions (eg not x, its y). bring the confidence level back to the one in the chapter”
That was MUCH better and required little revision. However, the drafted ended abruptly. I knew the right thing to do was ask people to reach out, but i felt like a jerk: everything sounded salesy. So I asked in the same thread, “can you write a gentle, humorous CTA to reach out to me if they need consulting or training to see if I am the right fit? I want to be clear that I i’ll tell them if I am not. Please draft 7 options.” The funny ones were not funny, so I went with the simple one and added the “over the skis” line myself, because it felt evocative and sincere.
