We asked AI to show us the secret aircraft hidden in Area 51

The world seems to be going through a sort of artificial intelligence gold rush in recent months, with generative AI tools like ChatGPT and DALL-E changing the way companies look at repetitive and even creative tasks across practically every industry. As a writer and content creator, I’ll be the first to admit that I looked upon this ongoing AI revolution with a combination of childish excitement and looming concern. After all, as cool as these new AI-powered gadgets may seem, writers like me may be among the first to lose their jobs to the comparatively lower costs of training an AI agent to relay the news.

It’s important to remember, however, that this is not a new debate. Today’s content creators may reign on platforms like YouTube and TikTok, but content creation is a job as old as time – and this is hardly the first technological revolution it’s weathered. When Gutenberg’s printing press emerged, religious leaders described it as the word of God being “spread about like dung,” rather than being painstakingly copied by hand. When e-books hit the market in the early 2000s, a panic struck the publishing industry as printers and authors alike lamented a new age of literary piracy that, thus far, has still yet to fully manifest.

This is to say that new technologies are often met with consternation among those they will affect most directly. And here in the second decade of the 21st century, AI is that new technology… and I have taken my place among the anxious luddites of history, worrying that our value to the world will diminish as automation steps in to streamline creation.

So, while we’ve long used AI for support functions like upscaling historical footage or improving the sound quality of recordings, last weekend, I decided to take a swing at going even further: Leveraging generative AI to produce original images that, themselves, are interesting enough to become the content I create.

And it was with that in mind that I gave the OpenAI-based DALL-E 2 15 of my hard-earned dollars to create some unique and unusual aviation-centric images, starting with a simple prompt: “Black and white photographs of Top Secret stealth aircraft hidden in Area 51 with men in suits and military uniforms looking on from the side.”

Related: Why today’s military AI isn’t capable of ‘going rogue’

The ‘Top Secret Aircraft at Area 51’ DALL-E 2’s Generative AI created

While some of the prompts I gave DALL-E 2 resulted in some pretty rough results (the AI seems to struggle to create images of existing aircraft), I was pretty impressed by the results for this prompt about Area 51. At first glance, these images have a haunting realism to them – even if that realism is fleeting once you look closely at the people in the image.

It’s interesting to note the scale of these platforms. I provided the AI with no requirements about the size of these top secret craft, but DALL-E 2 certainly seems to think bigger is better.

It’s also hard to deny how alien-looking these craft are. I have to assume that’s based on DALL-E 2’s contextual understanding of Area 51 as a location that the conspiracy-minded contend houses secretly recovered and reverse-engineered alien craft. Of course, that may be the result of my projecting context onto the AI – and I’d be interested to hear from those with sufficient technical expertise to confirm or dismiss that assumption.

In this image, DALL-E 2 clearly continues to struggle with creating realistic people, but the craft present in the background does look a bit more earthly in design… Almost like a secret stealth space shuttle, perhaps?

It’s honestly tough to get past just how eerie these photos look, especially when viewed on smaller screens like my phone. Despite the unrealistic depictions of men in suits, these pictures do creep into the depths of uncanny valley for me.

I saved the above for last simply because it appears the most alien to me, and could easily be misconstrued as a leaked photograph of a reverse-engineered flying saucer if you don’t look too closely at the people.

It’s important to understand that these images were the first pass at my prompt’s from DALL-E 2’s generative AI, and each image can be modified and improved upon for better results. That is to say that, while these images are pretty immediately recognizable as fake, it could be entirely feasible to refine them into far more realistic renders.

Likewise, with minimal expertise in graphic editors like Adobe Photoshop, even these unrealistic renders can be made more believable without a great deal of work. And that really begs the question… what could someone create if they were willing to invest more than 20 minutes and a few bucks?

Chances are… the result would be pretty believable, even if that person had extremely limited technical expertise.

Read more from Sandboxx News

• Hell in the Highlands: The men infiltrate the Korean camp
• Marines are training to provide security for commercial ships in Middle East
• Iliff Richardson’s efforts against the Japanese in the Philippines were so impressive that he earned medals from both the Army and Navy
• This former Marine Cobra pilot saw combat in Afghanistan. Now, she’s commanding her first space mission
• Operation Acid Gambit: Panama, Noriega, and a CIA man