The Tomb Raider AI Shortcut: Why 'Getting to Right Answers Faster' Is a Lie
Crystal Dynamics uses generative AI to visualize game ideas before building them. But this 'shortcut' kills experimentation and replaces human iteration with the average of everything already made. We explore why this mindset is dangerous for creativity—and how tools like ccLuca prove that AI can be used for genuine productivity without sacrificing originality.
There is a sickness creeping into the creative industries. It wears a friendly mask: efficiency. It promises to save time, to reduce costs, to help you "get to right answers faster." But what it really does is replace the messy, beautiful process of trial and error with a sterile regurgitation of the past.
Crystal Dynamics, the studio behind Tomb Raider: Legacy of Atlantis, has admitted to using generative AI to visualize game concepts before committing to a traditional pipeline. Their experience director, Jeff Adams, told Game Informer:
"At Crystal Dynamics, we see AI as a tool that can help our team get to right answers faster."
Faster. Always faster. But to what end?
The Average of All Inputs
Let me be clear: generative AI, as it exists today, cannot innovate. It can only compute the statistical average of everything it has been fed. It is a mirror reflecting a thousand other mirrors. It produces the mean, the median, the safe.
Adams explains that the team generates an idea using AI, checks if it "works," and then moves it to the traditional pipeline for human crafting. But here is the question he does not answer: How do you know it works if you never build it?
A visualization is not a prototype. A concept art generated by a machine is not a playable level. You cannot test a ghost. You cannot iterate on a hallucination.
Experimentation Has Been Thrown Out the Window
The real value of game development—of any creative work—lies in the failures. The dead ends. The prototypes that break and teach you something unexpected. That is where innovation lives.
By using AI to pre-filter ideas, Crystal Dynamics is effectively saying: "We only want to pursue ideas that look like things we already know work." This is not creativity. This is risk aversion dressed in a digital trench coat.
Adams concludes: "We just want to make it as easy as possible for us to make high-quality game experiences."
Easy. There is the word again. Nothing good is ever easy. The best art, the best code, the best expense reports—they come from struggle, from iteration, from the human willingness to be wrong.
Where AI Actually Helps (Without Killing the Soul)
I am not a Luddite. I do not believe all AI is evil. But I believe in using the right tool for the right job.
Consider expense tracking. It is tedious. It is repetitive. It is a perfect candidate for automation—not because we want to eliminate human judgment, but because we want to free humans to focus on things that matter.
That is why I built ccLuca.
No generative visualization of fake receipts. No AI pretending to be creative. Just a tool that extracts data from a photo in three seconds, generates reports instantly, and gets out of your way. It does not replace your thinking. It removes the friction so you can think about something else.
The Difference Between Shortcut and Leverage
A shortcut bypasses the work. Leverage amplifies it.
Crystal Dynamics is using AI as a shortcut. They are skipping the hard part—the building, the testing, the failing—and hoping the machine can predict what will succeed. It cannot. The machine only knows what has already succeeded.
ccLuca, on the other hand, is leverage. It does not tell you what expenses to claim. It does not generate fake receipts. It simply makes the process of capturing and reporting your real expenses faster. The human still decides. The human still owns the outcome.
The Real Cost of 'Easier'
When you outsource your thinking to a machine, you lose something intangible but essential: the ability to be surprised.
Video games used to be great experiments. They redefined technology, storytelling, and interaction. Now, many studios are content to remix the past with the help of a statistical parrot. The result is polished, competent, and utterly forgettable.
We are building a world where everything is average. Where nothing offends, nothing challenges, and nothing surprises. That is not progress. That is a slow, comfortable descent into mediocrity.
A Better Path
Use AI for what it is good at: pattern recognition, data extraction, repetitive tasks. Use it to handle the boring parts of your life so you can spend your energy on the parts that require imagination, empathy, and courage.
That is what ccLuca does for expenses. It handles the drudgery. You handle the rest.
Crystal Dynamics could learn something from this philosophy. Instead of using AI to pre-chew ideas, they could use it to automate the tedious parts of development—asset management, bug tracking, build testing—and leave the creative work to humans.
But that would require admitting that the creative work is the point. And in 2026, under the weight of new-age capitalism, that seems to be a luxury most studios cannot afford.
Source: Tomb Raider: Legacy of Atlantis devs use generative AI to 'get to right answers...'