The Technology Was Never the Mission — Remembering Them Is

Every now and then, I come across something online that reminds me why we try so hard to do this work the right way. Recently, it was an AI reconstruction of a photograph of Ernest Elijah Polston, one of the 134 men killed aboard USS Forrestal on July 29, 1967. The image missed the mark completely. Somewhere underneath that reconstruction was a real man, a real sailor, and a real life, yet what had been created barely seemed to represent him at all.

My immediate thought was simple: Ernest Polston deserves better than that.

That also made me realize that there is something important we probably need to explain about what we do here at Ghosts of the Battlefield. We use artificial intelligence. We do not hide that fact, and we are not ashamed of it. But there is a tremendous difference between using AI to invent history and using AI as a tool to help preserve, study, and better understand the history that already exists.

Most of the time, we begin with an actual historical photograph. It may be faded after seventy or eighty years. It may be scratched, torn, stained, badly exposed, blurred, or damaged so severely that some of the original detail has almost disappeared. We use modern restoration tools, including AI, to help repair that damage, clean up deterioration, improve contrast, and recover details that have become difficult to see.

Then, in many cases, we work to colorize the photograph. To some people, colorization may seem like nothing more than making an old photograph look more attractive. For us, it can be much more than that. One of the things I have noticed time and time again is that color can cause details to suddenly jump out of an image that had almost disappeared into the shades of gray. A piece of equipment becomes easier to separate from the uniform behind it. Different fabrics become more obvious. Insignia, web gear, vehicles, terrain, buildings, aircraft, and even small details in clothing can suddenly become easier to recognize.

The world those men lived in was not black and white just because the photographs were.

I learned that lesson long before artificial intelligence entered the picture. When I first became involved in Civil War reenacting, there was a general belief among a lot of people that everyone during the period dressed in dark, dull colors. You looked at those old photographs and saw blacks, grays, whites, and endless shades between them, so it was natural to imagine that the world itself must have looked much the same.

But it didn't.

Over the years, as people studied surviving garments, fabrics, dyes, period descriptions, military specifications, paintings, and other evidence, it became obvious that many of those supposedly black or gray clothes had actually been very different and sometimes very vibrant colors. The photographic processes of the nineteenth century did not reproduce every color the way the human eye saw it. Two completely different colors might appear almost identical in a photograph, while another could appear far lighter or darker than we would expect today.

That lesson has always stayed with me.

Today, AI gives us another tool for exploring those same questions. It can help recognize visual patterns, assist with damaged areas, compare possibilities, and help us investigate what certain colors, equipment, materials, or details may have been based on a tremendous amount of available information. And before someone says, "Well, then you aren't doing the research," I understand the argument. In a sense, they are right.  AI can now help us search and compare an amount of information that one person could never realistically work through on their own. But that does not somehow make the research meaningless. The point of research was never to prove how difficult we could make the process. The point was always to learn.

Those of us who grew up before the internet probably remember what research once looked like. You used whatever books were available at the local library. Maybe you had a set of encyclopedias at home. You went through old magazines, reference books, photocopies, microfilm, and whatever else you could find. At the time, many of those sources felt absolute.

Sometimes they were excellent. Sometimes they were wrong. Sometimes twenty years later, new evidence completely changed something we had believed for years. We were never perfect researchers before AI.

AI is not perfect either. That is why oversight still matters. We have to know enough about what we are looking at to recognize when something simply does not make sense. AI can make mistakes. It can misidentify something, choose an incorrect color, invent a detail where damage exists, or produce something that looks convincing while being completely wrong. When we see that, we correct it. Sometimes we start over. Sometimes we leave something alone because we simply do not know enough to say what was there.

The tool has changed. The responsibility has not.

There is also another reason colorization matters to us, and it may be one of the most important.

We live in a world of color photography, high-definition video, and images that surround us constantly. For younger generations especially, black-and-white photography can unintentionally create distance. A photograph taken seventy or eighty years ago can feel as though it belongs to another universe entirely.

Then you carefully add color to the photograph of a nineteen-year-old serviceman and something changes. The uniform becomes cloth instead of gray. His eyes become more noticeable. The aircraft behind him looks like a machine that could still be sitting on a flight line. The jungle becomes green. The mud becomes brown. The blood, unfortunately, becomes red.

And suddenly the person in that photograph feels a little less distant. He is no longer simply a figure in an old history book. He was alive. He laughed. He worried. He got tired. He wrote home. He had people who loved him. He probably wondered what he was going to do when all of this was over.

Some came home. Some did not.

That connection is one of the biggest reasons we do this.

When we restore one of these photographs, the technology is not the story. They are.

That is also why seeing that reconstruction of Ernest Elijah Polston bothered me so much. It was not simply that the AI had done a poor job. It was that a real man had effectively disappeared underneath what the technology had created.

Ernest Polston was born on March 10, 1936, and came from Lancaster, South Carolina. He served in the United States Navy aboard USS Forrestal with VF-11 as an Aviation Structural Mechanic Hydraulics Second Class. That was skilled work. Important work. The kind of job that helped keep naval aviation functioning in the middle of war.

On July 29, 1967, fire and explosions tore across Forrestal and claimed the lives of 134 men.

Ernest Elijah Polston was one of them.

He was not an AI reconstruction. He was not a piece of content. He was not simply another name among 134 casualties.

He was 31 years old. He was a sailor from Lancaster, South Carolina. A Navy man. A skilled mechanic. A human being who had already lived 31 years of a life that should have had many more ahead of it.

That is what we can never allow the technology to obscure.

Ghosts of the Battlefield was not created so that we could capitalize on the memory of men and women who served, and especially not those who gave everything. We are here because those people deserve to be remembered. Some gave years of their lives. Some came home carrying wounds. Some came home carrying memories they rarely spoke about. And some never came home at all.

If artificial intelligence can help us repair a damaged photograph, reveal a detail, understand something we could not see before, or help another generation look into the face of someone who served decades before they were born and feel even a little closer to that person, then I believe it is a tool worth using.

But it has to remain exactly that.

A tool.

It should never become more important than the history. It should never replace judgment. And it should never replace the human being whose photograph we are trying to preserve.

We are not trying to replace the historical record. We are trying to preserve it, study it, and make sure the human beings inside that record are not forgotten.

Because in the end, whether the work was done with an encyclopedia, a microfilm reader, a computer database, Photoshop, or artificial intelligence, the technology was never the mission.

Remembering them is.

Support the preservation of America's military history.