Concerns Regarding AI

While it is easy to get pulled into the excitement of new technology, it is also important to understand the larger context of that technology. The benefits of utilizing AI are vast, but so are the concerns. The exact impact of AI is unknown and fluctuates rapidly. We encourage you to consult more recent data when considering specific impacts, but a general overview of some of the concerns is provided below.

Ethical

AI is trained on the works of others. During development, it is exposed to massive amounts of existing content, but developers are often vague about what material is included in the training data. It’s unclear whether copyrighted content is part of this data and whether consent was obtained from the original creators—most likely, it was not. While we have frameworks for resolving intellectual property disputes between people, AI complicates these issues in ethically gray and unprecedented ways.

Similarly, the lack of transparency regarding training data can result in sensitive information becoming public that perhaps should remain private. One concern is that user data may be fed back into training systems. For instance, if you use AI as a financial advisor and input personal information into the chat, there is a risk that this data could be included in future training sets and inadvertently shared with millions of other users, without your consent. You should never enter personal information into an AI program.

AI lacks the capacity to feel emotions or understand the consequences of sharing sensitive information or offering poor advice. Despite its ability to communicate in human-like language, it is not human. AI operates on probabilities and has no grasp of morality or the repercussions of its suggestions. Therefore, relying on AI for spiritual guidance or wellness advice is strongly discouraged. AI does not fully comprehend the human experience and should not be a substitute for human insight on related topics.

Accuracy

AIs know a lot, but they don’t know everything and they are very bad at knowing what they don’t know. When a gap in training occurs, AIs can hallucinate or just make stuff up. The problem is, that they will not flag the information they made up based on nothing as different from the content they generated from reliable sources. In many cases, the AI may even try and convince you the erroneous material is correct.

AI’s reliance on training data also means that the systems can quickly fall out of date without frequent updates. Information has to be fed into the system to have it be kept up to date.

Bias

The AI is also only as informed as the data it was trained on. When considering all of the works of human history, many voices and perspectives have been left out of the preserved narrative. This means that the content generated by AI is normally slanted toward Western male views because that is what we have the most data on. This can lead to biased responses and narrow perspectives.

Environmental

Training and maintaining AI systems demand vast resources. The storage systems housing AI’s knowledge bases are enormous, requiring specialized chips that involve the mining of heavy metals and other finite resources. Rapid advancements in technology result in frequent hardware obsolescence, generating significant e-waste.

AI data centers also require constant cooling, consuming billions of gallons of water worldwide—including in areas already facing water scarcity. Additionally, these centers place immense strain on electrical grids, as they must remain powered and cooled at all times. The carbon footprint of these systems is substantial and continues to grow as more comprehensive models are developed.

The Future

AI raises other significant concerns, including its impact on job markets, financial costs, and the arts. However, while these challenges are real and pressing, AI is also being used to address some of the very same problems it is contributing to. For example, despite its large carbon footprint, AI can improve efficiency and sustainability across various systems, thereby reducing emissions in other areas.

The AI landscape is still evolving, and the full implications of this technology remain unclear. By addressing these concerns now, we can help shape AI into a versatile tool that benefits society.