25 Aug From Access to Authentication: What Nashville Taught Me About AI and the Future of Investigations

I went to Nashville recently to speak at NCISS on a panel about technology, artificial intelligence, and the future of investigations.

I came home thinking about musicians.

If you’ve spent any time in Nashville, you know they’re everywhere. Music pours out of nearly every bar, and what struck me wasn’t simply the quantity. It was how good so many of these people were. I heard singers with incredible voices and watched musicians who had obviously spent thousands of hours mastering their craft. In almost any other town, some of them would immediately stand out.

In Nashville, there were hundreds of them, many playing for tips.

At first, I was amazed. After a while, I found it strangely depressing. These were people who had spent years becoming exceptional at something, surrounded by hundreds of other people who had done exactly the same thing.

The problem wasn’t a lack of talent. It was an abundance of talent.

That thought stayed with me because earlier at the conference someone had asked our panel the question everyone eventually asks:

Is AI going to take our jobs?

I gave the comfortable answer.

Take part of our jobs? Probably. Replace private investigators? Absolutely not. There isn’t going to be an AI bot sitting in a witness chair, raising its right hand and explaining an investigation to a judge and jury.

It was a good panel answer.

On the way home, I realized it wasn’t a very good answer.

AI doesn’t have to replace private investigators. It only has to make some of what we do abundant.

When Capability Becomes Cheap

Testimony is a small part of what investigators actually get paid to do. A tremendous amount of our time goes into reviewing records, conducting background research, searching public information, examining social media, constructing timelines, comparing documents, reviewing video, cross-referencing data, and preparing reports.

Those are precisely the kinds of tasks AI is becoming increasingly capable of assisting with.

Take a 400-page document production. An investigator might spend hours extracting names and dates, identifying inconsistencies, building a timeline, and deciding what deserves further investigation. AI can increasingly assist with that first pass in a fraction of the time.

The same thing is happening with video review, public-record research, social media, document comparison, and large datasets.

The work doesn’t necessarily disappear.

It may simply stop taking three people and start taking one.

If you’re that one investigator, AI could be one of the greatest productivity tools of your career. If you’re one of the other two, hearing that “AI isn’t replacing investigators” isn’t particularly reassuring.

For investigative business owners, there’s another problem hiding inside that efficiency. We sell a lot of time. If something that historically required eight hours can now be responsibly accomplished in three, that’s an incredible productivity improvement—and potentially five fewer billable hours.

The investigation didn’t become less valuable.

The relationship between time and value changed.

That may eventually force our profession to answer a much bigger question than whether AI is taking jobs:

What are clients actually paying us for?

From Access to Authentication

For most of my career, part of an investigator’s value came from access to information.

We knew where to look. We had specialized databases. We understood public records and which agencies maintained them. We knew how to locate people, connect records, and find information that wasn’t obvious to everyone else.

Information was harder to obtain, so access had value.

I think that model is changing.

We’re moving toward a world where the biggest problem won’t necessarily be finding information. It will be determining which information is real.

AI can search enormous amounts of material, summarize records, identify relationships, construct timelines, and compare documents faster than any individual investigator. At the same time, the same technology can generate a photograph, clone a voice, manufacture a screenshot, alter a document, fabricate a text conversation, or create a convincing online identity.

We may soon have more access to information than at any point in human history while having less certainty about whether that information is authentic.

That changes the investigator’s value proposition.

The question used to be:

Can you find it?

Increasingly, the question will be:

Can you prove it’s real?

And after that:

Can you establish where it came from?

Finding a photograph may become easy. Establishing its provenance, locating the original file, examining available metadata, identifying the source when possible, looking for manipulation, finding independent corroboration, and explaining what the evidence actually supports is something entirely different.

For years, investigators have been information hunters.

I think we’re increasingly going to become information authenticators.

Verification Becomes the Job

This is where my original courtroom answer begins to fall apart.

It’s true that AI can’t sit in a witness chair. But technology already plays an enormous role in evidence. Digital forensic extractions, license plate reader records, cellphone data, analytical software, databases, and machine-generated information are already part of modern investigations and litigation.

The exact evidentiary requirements vary by jurisdiction and circumstance, but the larger point is that a human doesn’t necessarily need to have manually performed every technical step.

So the courtroom isn’t the moat I originally made it out to be.

The important question is who can responsibly stand behind the work.

Who knows where the information came from? Who verified it? Who understands the limitations? Who knows what the evidence proves—and what it doesn’t? Who can explain the methodology when challenged?

That brings us to something AI cannot independently assume:

Accountability.

An investigator has a name and professional reputation. There may be a license behind that name, an E&O policy, a signature on a report, personal knowledge, and eventually testimony under oath.

The model doesn’t independently take responsibility for the conclusion.

The investigator does.

That’s also why I don’t think we should build our future around saying AI makes mistakes. The technology is improving too quickly for that.

In fact, I think there’s a strange danger in AI becoming more accurate.

A tool that’s wrong 30% of the time teaches you to check everything. A tool that’s wrong 2% of the time teaches you to stop checking. Eventually that 2% gets into a report, an affidavit, or testimony because the human using the tool has learned to trust it.

I think of this as the Reliability Paradox:

The more reliable AI becomes, the easier it becomes to trust the rare answer that is wrong.

So the durable argument isn’t that AI is unreliable.

It’s that verification is the job.

Investigators should already understand this. Digital tools have been producing results that require interpretation long before generative AI existed. A timestamp can be misunderstood. A database connection can imply a relationship that isn’t there. An analytical tool can identify a pattern without establishing what caused it.

Computer output isn’t automatically fact.

The tool tells you where to look.

You still have to go look.

But What Did You Give the AI?

There is another question that may prove just as important as whether the AI’s answer was correct:

What did you give the AI to get the answer?

Imagine an attorney sends an investigator sensitive litigation materials. The investigator uploads them to an AI platform and asks it to build a timeline, identify inconsistencies, and suggest areas for further investigation.

The system produces an excellent analysis and saves several hours.

But case information may also have just been transmitted to an outside technology provider.

Now the questions change. Which platform was used? What information was uploaded? How is it retained? Who can access it? Can it be used to improve a model? What contractual protections exist? Do the client’s instructions permit the use? Does a protective order restrict it? Could the AI interaction later become relevant to a privilege, work-product, preservation, or discovery dispute?

Those aren’t simply IT questions.

They’re potentially questions of confidentiality, privacy, privilege, work product, discovery, preservation, and professional responsibility.

And those concepts aren’t interchangeable.

A system can be secure without making everything entered into it privileged. An enterprise account can offer substantially stronger contractual and privacy protections than a consumer account without making its contents automatically immune from discovery. At the same time, using a third-party AI provider doesn’t automatically waive work-product protection.

Recent federal cases have already reached different conclusions about AI interactions and work-product protection based on the circumstances. The law is developing, which is exactly why investigators should avoid broad assumptions in either direction.

Being retained by an attorney also doesn’t automatically make every investigative communication, note, report, prompt, or AI interaction privileged. And sometimes the answer is simpler: a client’s instructions, confidentiality agreement, engagement terms, or court-issued protective order may restrict what can be uploaded in the first place.

The fact that technology allows us to upload something doesn’t mean we’re authorized to do it.

AI also adds another possible layer to the investigative process. Instead of evidence moving directly from investigator to report, it may now move through an AI interaction before the investigator verifies the result.

That doesn’t mean every prompt is discoverable. It does mean AI use itself could someday become relevant to how an investigation was conducted.

I can easily imagine a deposition asking: Which AI platform did you use? What did you upload? Did it include confidential material? What did the system tell you? What did you rely on? What did you reject? How did you verify the result? Can you produce the underlying evidence?

Those aren’t questions an investigator should consider for the first time while sitting in the witness chair.

AI Is Also Creating the Next Investigation

If we only ask what AI will take away from investigators, we’re missing half the story.

AI isn’t just changing investigations.

It’s changing evidence.

Synthetic audio, deepfake video, AI-generated photographs, fabricated screenshots, manufactured conversations, altered documents, and synthetic identities are creating problems that barely existed in their current form a few years ago.

That changes the assignment.

We used to be asked:

What does this photograph show?

Increasingly, we may be asked:

Did this ever happen?

Think about that question in a custody dispute, an employment matter, an insurance claim, a fraud investigation, a harassment allegation, or civil litigation.

Someone may need to locate the original file, examine metadata, identify the source device when possible, establish provenance, compare versions, find corroborating records, interview witnesses, preserve evidence, document chain of custody, and explain what can—and cannot—actually be established.

That’s not investigative work being destroyed by AI.

That’s investigative work being created by AI.

And it brings us directly back to the shift from access to authentication.

I’d rather build that practice than spend the next decade defending an old job description.

What Nashville Actually Taught Me

I’ve thought a lot about those musicians since I came home.

The lesson wasn’t that musicians aren’t valuable. It was almost exactly the opposite. They were so talented that talent alone wasn’t enough to distinguish them.

That’s what investigators need to understand about AI.

AI doesn’t have to become a better private investigator than you. It doesn’t have to conduct surveillance, sit across from a difficult witness, or testify. It doesn’t even have to be right every time.

It only has to make certain capabilities that were once expensive and scarce cheap and abundant.

And when something becomes abundant, the market places greater value on whatever remains scarce.

For most of my career, investigators were paid in part because we knew how to find information other people couldn’t.

I believe the next generation of investigators will increasingly be paid because they can establish which information deserves to be believed.

That means judgment matters. Provenance matters. Verification matters. Reputation matters. Knowing the limitations of your evidence matters. Being willing to say I don’t know when the evidence doesn’t support an answer matters.

And accountability matters.

None of that means we should resist AI. I use it. Investigators who refuse to use these tools may eventually find themselves slower and less competitive than investigators who learn to use them responsibly.

But using AI and trusting AI are very different things.

My own standard is simple: AI-assisted analysis is a lead until I verify the material conclusion. When AI analyzes existing evidence, I return to the underlying source. Sensitive information doesn’t go into a system merely because doing so is convenient. I want to understand the platform I’m using, the restrictions governing the case, and what happens to the information I provide. And if I rely on a conclusion, I want to be able to explain how I verified it.

Most importantly, AI doesn’t change one of the oldest rules in investigations:

When the evidence doesn’t support an answer, say I don’t know.

Our job isn’t to manufacture certainty.

It’s to accurately report what the evidence supports.

So if someone asks me at the next conference whether AI is going to take our jobs, I won’t give the comfortable answer.

Some jobs probably will disappear. Some investigative hours almost certainly will. Some services will become commoditized, and some firms will have to rethink how they price and deliver their work.

But I don’t think the investigative profession is disappearing.

I think its value is moving.

From access to authentication.

For most of our history, investigative value was closely tied to finding information. In the world we’re entering, information itself may become almost impossibly abundant.

The future isn’t simply about having access to more information.

It’s about knowing what information can be trusted.

The investigator of the future won’t necessarily be the person who can find information fastest. Machines may win that race. It will be the person who can examine what the technology found, return to the underlying evidence, establish where it came from, understand its limitations, determine what it actually supports, and stand behind the conclusion.

Use the machine for speed.

Use the investigator for authentication, verification, judgment, and accountability.

The technology got faster. The information became abundant.

Truth didn’t.

And that may be where the future of our profession begins.

About the Author

Justin D. Hodson, CPI, is a licensed private investigator and founder and president of Hodson PI, a professional investigations firm he founded in 2003. With more than 25 years in the investigative profession, Hodson leads a team of more than 80 professionals providing surveillance, SIU, background, OSINT, digital evidence, and litigation-support investigations.

A recognized industry speaker and advocate for responsible technology adoption, Hodson regularly addresses the intersection of artificial intelligence, investigative methodology, privacy, evidence, and professional accountability. He was named the 2024 CALI Investigator of the Year and has been featured in PI Magazine.

His focus is not simply on how AI can make investigators faster, but on how the profession can use emerging technology without compromising verification, defensibility, confidentiality, and human accountability.

Disclaimer

This article reflects the author’s observations about emerging issues involving artificial intelligence and investigative practice and is intended for general educational and professional discussion only. It is not legal advice. Laws, court rules, protective orders, contractual obligations, and professional requirements concerning privilege, work-product protection, confidentiality, discovery, preservation, privacy, cybersecurity, and admissibility vary by jurisdiction and circumstance and continue to develop. Investigators should follow applicable laws, licensing requirements, client instructions, and court orders and consult qualified counsel regarding the requirements applicable to a particular matter.