Pope Leo XIV released his first encyclical two months ago. The letter, which focused mostly on artificial intelligence (AI) and the potential risks the technology poses to humanity, promptly garnered global attention.

The concerns the pontiff raised include mass displacement of workers by AI, exposure of children to violent, hypersexual information online, the use of automated systems in warfare, and enormous concentration of digital power in the hands of a few players.

Christopher Olah, the co-founder of Anthropic, which is one of the companies at the forefront of the AI revolution, was invited by Pope Leo to the presentation of the encyclical at the Vatican. In his remarks, Olah welcomed the pontiff’s voice in the AI debate. He said companies like his own need moral guidance to avoid being swayed by “a set of incentives and constraints that can sometimes conflict with doing the right thing.” Olah added, “We need moral voices that the incentives cannot bend.”

The Wall Street Journal says, “Anthropic has leaned in to philosophical questions such as whether AI models experience consciousness,” adding, “[the company] employs an in-house philosopher to help instill morality in its AI.”

By going as far as hiring a philosopher to train its AI models, Anthropic was clearly demonstrating that it takes the technology’s risks seriously, even before Pope Leo issued his encyclical. While that is laudable, the idea of humans teaching AI systems to act ethically is quite amusing.

The Pew Research Center reports that 75.8 percent of the world’s population identified with a religion as of 2020. Religious teachings focus heavily on morality. Starting as early as kindergarten, we are taught the proper ways to relate to our fellow human beings. Our parents and guardians try to impart ethical values to us over many years at home. So, whether we are theists or atheists, we all have ample opportunities to learn to distinguish right from wrong. And yet, the world is full of sin.

Even the most casual look around the world today quickly reveals a deeply unflattering picture of human nature. Unadulterated greed is on display everywhere. There are senseless killings of thousands of innocent people happening almost daily. There is mindless looting of state resources that immiserate too many people in too many places. The list of our vices is embarrassingly long. We are either exceptionally poor learners, or totally incapable of putting into practice the virtues we are taught.

If we are such disappointingly flawed beings, what makes us think we have the capacity to instill morality into any AI systems we design to make them virtuous? There is a hint of arrogance in that thinking. Are we simply going to tell the systems to “do as we say, not as we do”?

Genesis 1:27 says, “God created man in His own image.” I wonder what happened then, after He placed us on Earth. Whenever I ask that question, the answer offered is that Satan finds ways to get to us, and is often successful at steering us away from the virtuous path God charted for us. If that is the case, how much luck do we think we will have in our own efforts to keep the demons away from our AI creations? I can’t help but think that God is watching us from His perch and laughing at our arrogance.

I hope Anthropic and the other companies that are currently examining whether AI systems experience consciousness will soon have a clear answer to that question. If it turns out that these systems do indeed have that human feature, my next prayer will be for the systems to have a higher level of cognizance than we do. It would be a disaster if our digital creations assumed our character. The last thing we want to do is enlarge the pool of bad actors in the world.