What do you think are the positives and negatives of AI? Normally, people might come up with pros such as doing chores for us especially in Japan with the problems with the overpopulation of the elderly. This is true and it is a big positive aspect of AI, but I feel like we do not realize the simpler pros of AI itself. In this blog, I would like to give you some insight on the basic advantages and disadvantages of AI and talk a bit more about the dangers of it, as we care more about if AI will one day “take over the world”. I hope you can gain something from it!
< Advantages >
Khanzode and Sarode of IAEME (The International Association of Engineering and Management Education) state that AI's main advantage is that it can finish difficult tasks faster, efficiently, and with accuracy. They are likely to make fewer errors in fields that deeply involve algorithms (although AI itself is assembled from algorithms), and the possibility that their task will succeed is very high. AI basically combines large data with intelligent and repetitive mathematical computations as they process algorithms to learn patterns in the data they analyze. The strong thing about AI is that they do not need a break. If humans were to analyze the same kind of data, our efficiency and accuracy will highly likely drop as time passes because we obviously get tired from staring at our computer screens for too long.
Pew Research claims that “data-driven approaches to problem-solving will expand”, “code processes will be refined and improved; ethical issues are currently being worked out”, and in the near future, “the world may be governed by benevolent AI”. Although they praise AI for its visible and invisible benefits towards providing humanity with great insight, they mention many challenges and concerns about AI.
< Disadvantages >
Pew Research Center states there are five concerns and societal challenges of AI.
One is that when data and predictive modeling become more important, humanity and human judgment can be lost. It is not good to use programming as a tool to gain mainly profits and efficiencies as it can be a threat. The danger is that algorithms can "read our minds" and manipulate people for the outcomes. Manipulations will lead to a "flawed yet inescapable logic-driven society", as Pew Research Center explains. Some researchers are concerned about how AI could lead to us losing our sophisticated decision-making capabilities and local intelligence. If algorithms take over our society too much, we humans would become outliers. Pew Research claims that as a solution, we should embed into AI the ability to respect humans.
Another thing Pew Research Center claims is a concern is how even something that is seen as flawless like algorithms can be biased. "Algorithms depend on data that is often limited, deficient, or incorrect." If the programmers are incorrect or biased, AI would reflect those biases and imperfections.
The third concern is that AI's ability to categorize might lead up to "deeper divides". AI's categorizations will limit the number of ideas and reliable information people are exposed to and eliminate serendipity, as the information the user wants will only be picked up.
The unemployment problem is a widely known concern in AI. People and researchers are concerned that AI that is smarter than humans will replace their jobs and lead to a rise in unemployment rates. Due to this concern, some suggest that a "global economic system to support humanity" should be redefined.
Lastly, Pew Research states that people will need to understand algorithmic literacy, which is beyond the basic digital literacy that many schools nowadays teach. Additionally, people will be in demand for transparency and oversight, as many have a negative view of AI's policy rules and oversight.
We can see from these five concerns, that there is more to the cons of AI than just unemployment and AI eventually "taking over humanity". Daron Acemoglu, a professor of applied economics at the Massachusetts Institute of Technology states that what is more dangerous about AI is not AI itself, but how leading firms can approach data and use it. Unregulated AI can cause situations such as companies getting hold of excessive amounts of users' information. Acemoglu states that these "data externalities can contribute to too much data being concentrated in the hands of companies, with adverse implications for privacy and consumer surplus." What I thought was even scarier was how companies can use users' information to manipulate their behavior. Advertising is a form of manipulation in a way. Simple ads we see on our computer screens all the time may seem fine, but there have already been suspicious cases such as Target (an American department store chain) successfully predicting whether certain online Target users are pregnant women, and sending them secret ads of baby products. Even apps such as Youtube, Twitter, and even TikTok use their algorithms to send us more addicting videos or feeds to specific groups.
AI safety will be a big key factor in future AI development. We need to make sure that AI itself does not go the wrong way, and that humanity will also not use AI in an inappropriate manner.
This was our blog for question #5 and #8 of our 30 research questions.
References:
Acemoglu, D. (2021, November 23). Dangers of unregulated artificial intelligence. VOX, CEPR Policy Portal. Retrieved July 7, 2022, from https://voxeu.org/article/dangers-unregulated-artificial-intelligence
Colorado State University Global. (2021, August 9). How Does AI Actually Work? Retrieved July 7, 2022, from https://csuglobal.edu/blog/how-does-ai-actually-work
Khanzode, K., & Sarode, R. (2020). ADVANTAGES AND DISADVANTAGES OF ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING: A LITERATURE REVIEW. International Journal of Library & Information Science (IJLIS), 9(1), 34–35. https://doi.org/10.17605/OSF.IO/GV5T4
Rainie, L., & Anderson, J. (2020, August 6). Code-Dependent: Pros and Cons of the Algorithm Age. Pew Research Center: Internet, Science & Tech. Retrieved July 8, 2022, from https://www.pewresearch.org/internet/2017/02/08/code-dependent-pros-and-cons-of-the-algorithm-age/
Comments
Post a Comment