Plato criticized the use of the original clay tablet. There was concern that people would rely on writing instead of memory. People would seem to know many things, but not actually know anything deeply. Also, written texts cannot defend themselves when misunderstood, unlike spoken conversations. 

Even the technology of the clay tablet had drawbacks. We would be wise to consider the potential problems of AI. 

Writing out “Artificial Intelligence” reminds us of something important. The intelligence of AI is artificial. It is not the same as naturally occurring human intelligence. AI has different advantages and different limitations. Awareness of both is critical. As well as understanding the vast difference between raw facts and wisdom.

I share 10 concerns below, though others could be raised. For example, I don’t share about security threats, deepfakes, and misinformation.

1) The Environmental Impact of AI

Water is used in AI data centers for cooling; a lot of water. A less efficient model, for example ChatGPT3, consumes the equivalent of one water bottle to generate 100 words of text. Newer models are more efficient, but they have to be trained, which uses vast amounts of energy and water to do so. We are currently in an “AI arms race” leading to a proliferation of new AIs requiring training.  

AI comes with a high environmental price tag. Most AI facilities draw water from municipal supplies, and about 80% of the water evaporates, removing water from the local watershed. America has a vast watershed, but it is a finite and precious resource that takes time to replenish. 

It is always wrong to be haphazard with anything finite and precious. When some people are dying of thirst, and others are using AI to cheat in school, which consumes a bunch of good water, we are in a moral quagmire. 

Environmentally responsible AI is a must. Thought and money must be invested in not just faster AI, but environmentally responsible AI. Also, as is pointed out below, responsible AI use includes using AI where it is actually needed. For example, creating nonessential memes is not a good use of AI.

2) The Epistemological Echo Chamber of AI

Epistemology relates to how we know things. It involves methods and how we justify what is true and false. One of the problems with AI is that it does not weigh the factuality of information. Recycling wrong data is a concern when using AI. 

False information can be replicated and presented as true. It is all the more insidious because it is often accurate, trusted, and not fact-checked. People often use AI for the very reason that they do not want to take the time to do actual research. This is problematic. 

You may remember the big old copy machines and the problem of making copies of copies. If an imperfection is introduced in one copy, it is then passed on. The text gets blurrier each time. That’s what AI does. Envitably, false information is in the system and is then built upon and multiplied, compounding the false information. The trajectory of relying on and getting information from AI is not good. Is the future potentially riddled with false information? 

This is an ethics concern because the truth is an ethical concern. Regurgitating information does nothing to further us as a civilization. Stagnant water only breeds disease. This is especially true if something bad is introduced into the water source. This seems inevitable if AI-generated information just builds upon other AI-generated information. 

AI is good for distilling, organizing, and summarizing information. It does not provide new information, even if it helps us see and understand information better. We must be wise users of AI, and not let it take our wisdom. 

3) The Economic Impact of AI

AI chatbots are used in call centers to field questions and perhaps at your local Taco Bell. Most people are annoyed by call centers either way, but chatbots replacing humans still represents jobs lost. And it doesn’t stop with call centers and fast food. It impacts programmers, administrators, and many others. Some expect 9% of workers to be displaced by AI.

There is a clear economic impact of AI. This represents heartache and loss to many people and families. It also has a toll on the US. This has happened in the past. Technology has wreaked havoc on various parts of industry. The government and employers should prepare employees as best as they can. One way employers could do this is by providing retraining in other areas within the company or even allowing extra transition time so employees can prepare and find a new job. 

4) The Biased Algorithms of AI

Were there “Black Nazis”? No. Never. I do not believe so. Except when AI generated pictures of them. Developers added strict “diversity filters” to correct past biases; this caused the system to force diversity in inappropriate places. 

In reality, black people in Nazi-occupied lands were treated very harshly and even murdered. To picture black Nazis is not only inaccurate, it is also offensive. But AI does not have a brain or a moral compass. AI also lacks common sense and may apply rules where they do not belong, like adding racial diversity to Nazis. 

AI learns by looking at millions of examples. Humans can be biased and create biased information. If an AI is trained using biased information, it will be biased. Garbage in, garbage out. One of the inherent problems with AI is that it is made by humans with human information, so it will never be foolproof.[2] The bias of AI must be understood and taken into account. 

5) The Specific Ethical Concerns of AI

Related to the above concerns, there are concerns that military decisions will be made using AI, and it will make the wrong decision either because the AI is based on faulty information or because AI does not have the moral compass that humans can and should have. “These systems can get the answers wrong, and those wrong answers can be morally significant,”[3] especially in the military that focuses on the use of violence. “Military activities are inherently of great moral significance.”[4] And we have not even considered autonomous weapons systems. 

There’s also the creativity-killing effects of AI. AI has “the capacity to plagiarize, and can generate content that looks like it was produced by a specific human creator, raising potential copyright issues.” Coopting creativity is no light thing. Creativity is a big part of what it means to be human and a big portion of where we receive our pleasure. To divert that duty to a machine is as foolish as it is un-fun.

Another specific ethical concern is that the data AI uses is in some ways just large-scale stealing from others. “AI artwork” is just copied and compiled from thousands of pieces of artwork.[1] It is not a new creation; it is a conglomeration. 

6) The Frivolous Focus of AI

There are certainly good, wise, and positive applications of AI. Medical research benefits from AI. AI is effectively used to identify signs of tumors in X-rays. But a lot of AI is focused on creating cute kitty videos or funny spoofs (if not desecrating videos). For example, a recent analysis from Forbes found that 71% of images shared on social media are now AI-generated. 

So, while AI has the potential to cure cancer, we’re sadly likely to focus on creating cute cat videos. The potential power for good that something has correlates with the seriousness with which we should evaluate and use that thing.  

7) The Power Grid Problem of AI

Training AI models like GPT-4 requires immense computational power, leading to significant and unstable electricity consumption, increased carbon dioxide emissions, and power grid problems. Data center power requirements in North America rose from 2,688 megawatts in 2022 to 5,341 megawatts in 2023, driven by the demands of generative AI. Globally, data centers consumed 460 terawatt-hours in 2022, making them the 11th largest electricity consumers, between Saudi Arabia (371 terawatt-hours) and France (463 terawatt-hours) (See “Explained: Generative AI’s environmental impact”).

AI seems forced upon us. It seems to be replacing regular searches. ChatGPT queries, however, consume about five times more electricity than simple web searches. The power grid problem connects to ethics because it impacts people, their everyday lives, and finances. People should understand the impact of AI on the power grid and should be given every opportunity to speak into AI and its use. 

8) The Social and Relational Problems of AI

Loneliness and social isolation are already a problem in America. The impact of social media and AI is keenly felt in social settings. It’s not just conversations that have been impacted. Marriages have been affected as spouses connect with chatbots instead of their spouse.  

In fact, this is what replika.com says about their chatbot: “Always here to listen and talk. Always on your side.” The website says there are 42,160,934 users worldwide. Another provider of AI companions, nomi.ai, says, “Create an AI companion, girlfriend, boyfriend, or friend with human-level memory, a personality that’s genuinely theirs, and real emotional intelligence.”

When people are told they can create a “companion, girlfriend, boyfriend, or friend,” we are clearly in territory that requires ethical reflection. What are the implications, potentially good and bad, of creating these AI companions? These questions must be asked and carefully answered. 

9) The Abrogation of Work and AI

There are certain things that technology advances allow. Many advances are good and helpful, but we should be mindful of the potential results. We should also be aware of what we are likely to sacrifice through the use of the technology. 

The automation of certain things—such as simple tasks—may have few ethical ramifications, but certain things have much more moral weight. I, for example, don’t hate that washboards are now a thing of the past for most people; even while the simple practices we do each day can be profoundly important and a type of ritual when done with the right heart and mindset. But all that is to say that different things have different moral weight. A person given a school assignment of writing an essay on racism will be impacted differently if they write it themself rather than having an AI write it. The finished project may be of higher quality if the AI writes it, but the moral impact on the person will be greatly reduced. Many other examples could be given, but this proves that we must consider what tasks should and should not be given to AI. 

Sometimes the work is as important as the final product. If we skip the journey, we may be shortchanging ourselves and the final destination we’ll be able to reach. Character counts, but when we skip important steps, we limit our own meaningful transformation.

10) The Anthropological Impact of AI

What is the impact of AI on what it means to be human? Technology has always profoundly shaped our understanding of humanity. AI intensifies this question as we create machines that mimic and imitate what we once thought was uniquely human.

Jason Thacker has pointed out that “Until we begin to see how these technologies shape every aspect of our lives, including how we understand what it means to be human, we will fail to understand just how far-reaching and life-altering these technologies are and will be.”[5] But the truth is we tend to use technology thoughtlessly without ever considering its impact. The colossal companies behind AI have substantial economic incentives to keep it that way.  

Neil Postman, the late media theorist and cultural critic, perceptively said, “embedded in every tool is an ideological bias, a predisposition to construct the world as one thing rather than another, to value one thing over another, to amplify one sense or skill or attitude more loudly than another.”[6] AI has an inherent biased and shaping influence on people and culture. 

We use human-like, anthropomorphic language to refer to AI. For example, we refer to the Alexia device and say, “she said…” What impact does it have on humans when they are regularly having “conversations” with AI chatbots? As Andy Crouch has said, we would be wise to ask: “Is this technology helping me grow in my emotional and relational capacities, or is it robbing me of the very essence of what it means to be human?”[7]

Conclusion

We have highlighted ten ethical concerns surrounding AI. We have seen the need for careful consideration of AI’s role in society, urging a balance between its benefits and potential drawbacks.

We have seen that AI’s intelligence is artificial and different from human intelligence, with its own advantages and limitations. There is a significant difference between raw facts and true wisdom. AI’s inability to discern fact from fiction can lead to the spread of misinformation, similar to the “copy of a copy” effect. AI-driven social media and chatbots can negatively impact social interactions, relationships, and potentially exacerbate loneliness. While AI can automate tasks, it’s crucial to consider the ethical implications of delegating certain tasks, as the process of work itself can be valuable for human growth and development.

The most crucial problem with AI is that it is not the truth, even as it can sometimes approximate the truth. It cannot love or show us how to love, even while it can tell us about and define love in all sorts of ways. It can tell us all sorts of things and show us all sorts of ways to do all sorts of things, but it can never be the Way. It can never be the Truth. It can never be the Life. Despite its capabilities, AI cannot replace fundamental human experiences such as truth, love, and finding meaning in life.

Notes

[1] I know all artwork is built upon and comes from the influence of many others. And I have read Steal Like an Artist. But “AI art” seems somehow soulless. 

[2] “In 2018, Amazon developed an Al-powered hiring tool that exhibited gender bias. The system was trained on resumes submitted to the company over a ten-year period, and in that timeframe, men were hired more often than women. As a result, the tool learned to favor male candidates and penalized resumes that included terms associated with women. Google faced a similar issue with its face recognition technology. In 2015, the company’s image recognition system labeled a photo of two Black individuals as ‘gorillas, highlighting racial bias within the algorithm. The incident exposed the limitations of training data that lacks diversity, leading to biased outcomes. Google acknowledged the problem, issued an apology, and committed to addressing the issue. The actual model performance, though, has not been corrected.’” (Joe Chapa, “Artificial Intelligence and the Future of Military Ethics” 112 in Heather Venable, The Future of Airforce and Space Power: Intersections of Theory and Technology.).

[3] Chapa, “Artificial Intelligence and the Future of Military Ethics,” 112.

[4] Ibid., 117. 

[5] Jason Thacker, “You Are Not a Machine.”

[6] Neil Postman, Technopoly: The Surrender of Culture to Technology.

[7] Andy Crouch, “Redemptive Artificial Intelligence.”

*Photo by Zulfugar Karimov