Artificial Intelligence and its Forthcoming Regulations

Image source: Vincent Kilbride, for The Economist

Alice Huaut 

May 17th, 2023

When a new product or a new technology is introduced not everyone will have the same reaction to its arrival. The diverging perceptions towards a new kind of technology are also paired with the extent to which a technology can influence people’s lives and whether it requires the creation of a new set of regulations. For instance, as cars were invented and became more widely used, traffic regulations were created.

The European Parliament is moving closer to establishing the first rules on artificial intelligence (AI) with the AI Act. The draft was voted on Thursday, May 11th, 2023, and should be in use by 2024. Italy’s temporary ban on Chat GPT demonstrates the uneasiness felt by the rapid growth of AI. This article will first define Chat GPT, then outline some of its problems, and finally a brief explanation of the EU’s AI Act.

Chat GPT was created by the company Open AI, founded by Sam Altman. Chatbots can be defined as internet chat engines. These tools have existed in the past, the first was coined with the name Eliza and invented in 1994. However, Chat GPT has led to the active rethinking of our societal models as this chatbot has surpassed what chatbots have done in the past. It is wired to collect a wide range of information present on the internet stored in knowledge databases. Then, it feeds answers to questions people ask the chatbot. This chat engine can be understood as a more advanced feature of the autocomplete functions on one’s phone when suggested words appear in the section of the screen above the keyboard–while texting–. The artificial intelligence or autocomplete function allowing this chatbot to work is not perfect, the bot still struggles to respond to ‘real situation’ problems. However, Chat GPT can complete a wide range of tasks from writing songs, resumes, and essays, to programming and the list goes on. Furthermore, with its latest update, Chat GPT 4 can now complete a law school exam and score in the top 10%–when it previously used to score in the bottom 10%–. Chat GPT still presents weaknesses, university professors claim the top papers remain those not generated by AI, but for an average mark, Chat GPT can do the job.

As a result, artificial intelligence is already facilitating the work of many and will surely strip away jobs shortly. The growth and perfecting of AI will most likely stimulate a new phase of creative destruction. The term coined by Schumpeter in 1942 speaks of the situation where technological change will lead to the destruction of a set of jobs but create new jobs in the process. This shift in the economy’s dynamics is making people rethink how they will be able to contribute to the workforce in the years to come. Students pursuing computer science degrees are becoming increasingly aware of the importance of specialization in machine learning, AI, or databases as bots can increasingly program with greater accuracy.

However, the urgent issue the EU Parliament desires to tackle with the AI Act is directed toward chatbots’ poorly regulated use of individuals' personal information and diffusion of misinformation. Prohibiting the use of people’s personal information is of great concern to the EU as it could pose risks to people’s safety. AI can create a social scoring system where people can be categorized based on their socioeconomic status or personal characteristics such as race and gender. The MEPs have created a list to include bans on intrusive and discriminatory uses of AI. For instance, demanding the prohibition of biometric identification systems in public spaces, emotion recognition systems, and “the scraping of biometric information from social media or CCTV footage to create facial recognition databases”.

In addition to protecting people’s personal information, the AI Act will demand greater transparency measures to prevent the illegal use of copyright data. In addition, indicating whether an image or text is AI-generated will become a requirement as chatbots are not disposed of fact-checking properties. Chat GPT cannot differentiate an academic paper and a fictional short story, the same weight will be given to both. Just like chatbots, image bots allow individuals to type the idea of an image and use images online and the ability to rearrange pixels image bots–such as Midjourney or DALLE.E–can construct the desired images. Whether it might be through texts or images the use of information present on the internet by AI-fueled tools may contribute to the internet becoming a less vibrant place. For instance, individuals may fear that posting their work online may be used by a chatbot and the chatbot will be consulted over the work of the individual publishing their work online. The distinction between human and AI information would help limit the issues encountered presently with the rise of AI.

John Hewshow, senior director of search engine optimization at Vimeo, states that “society has the most to lose with conversational AI”. Whether one agrees or disagrees with this statement does not prevent AI from evolving as we speak. Some may see this technology as a neutralizer of critical thought. However, rather than a Manichean view on AI, being aware of its limitations and advantages can help mitigate its risks and perhaps help people come up with new solutions to its prevailing ethical and moral dilemma. “One way to ensure that A.I. is used as a force for creation—rather than destruction—is through education.” Just like calculators which are authorized for some exams and are restricted for others, AI needs to be viewed as a tool rather than a creature. How and why people will use this tool needs to be studied meticulously and regulated.