How Reliable is Artificial Intelligence?
How reliable is artificial intelligence?
The question has been raised by countless communication professionals, PR experts, and journalists: How often is AI wrong or misinformed? This question applies to AI text and image generators alike.
While I do not have the answer to this admittedly complex inquiry, the following thoughts are intended as a discussion starter.
How true are AI-generated text or photos?
To answer this question, I have experimented: I asked four AI text generators the same question, with the same prompt: “Write a short description of me.” I did this using the best-known AI programs, a list that includes OpenAI’s ChatGPT, Google’s Bard, Liner, and Microsoft Bing. The AI programs hallucinated, which means it generated responses that contained false or misleading information presented as fact.
For instance, these programs awarded me educational qualifications that I did not complete but did not list the ones that I completed. The programs attributed awards to me that I did not receive. The titles of my books were also incorrect. A short CV written by Bing was correct, as were the references it provided.
But I used different AI photo generators to create pictures of myself. Not a single picture looked like me. I spoke to a professor at Columbia University who was working on the subject and he confirmed my opinion. He posed the following question: "What will today's students in higher education learn if they are learning from AI misinformation?"
AI is a very recent development in the information technology industry. This is why I think that it is important for communication professionals to develop and agree on rules of AI and its usage.
Artificial Intelligence Ethics and Compliance
Recently, PR professionals in Europe have thought through the issue in the title and created a code of ethics for artificial intelligence. In Warsaw, the International Communications Consultancy Organisation (ICCO) accepted the Principles for Ethical Use of AI in Public Relations. The ICCO later published an article determing that:
“Public relations and communications professionals in PR agencies are increasingly utilizing AI technologies to enhance their work and deliver impactful results. However, as AI becomes more prevalent in the PR industry, it is crucial to establish guidelines that ensure AI’s ethical and responsible use.”
Additionally, the organization noted the following:
“PR professionals’ confidence is returning as usage of AI tools shoots up, according to the latest ICCO and PRCA Confidence Tracker published. Confidence is returning with 78% confident or very confident. This is up from May, when the figure had dipped to 73%, but is still lower than any reading in 2022 or 2021.”
Expectations About the Use of Artificial Intelligence
I would like to add a few thoughts to the Warsaw Principles, the ethical and compliance issues of AI from the journalist's perspective. I would consider the following ethical rules of the AI to be fundamental. It is crucial to establish and follow ethical guidelines to ensure responsible and compliant use of this technology.
It is important that the ethical principles establish that:
Text, photos, and illustrations generated by artificial intelligence should be subject to the same ethical rules as any other form of journalism.
Ethical rules for AI should prioritize the well-being and safety of individuals while promoting transparency and fairness.
Transparency in AI decision-making processes is key to understanding it effectively.
AI ethics should also encompass privacy concerns. All participants must ensure that AI systems are compliant with privacy regulations and properly handle sensitive user data.
Compliance with copyright laws is essential when it comes to content developed by artificial intelligence and that we clarify who owns the copyrights for AI-generated content.
All users need to be aware of intellectual property rights and ensure that their AI systems do not infringe upon these rights.
I would consider it important to control and to check the correctness of texts and illustrations generated by artificial intelligence. Just as a reminder: when Wikipedia started, it published unvetted and untrue information. Previously, as a university professor, I did not accept papers that referred to Wikipedia. All this was due to the lack of control at the time. Today, this situation has changed; articles appearing on Wikipedia are now controlled, and checked.
Adhering to ethics and compliance in AI is essential to foster responsible and accountable use of this technology. Establishing ethical and compliance rules and ensuring compliance with copyright laws are key steps for artificial intelligence and its results to be used correctly by all participants, content generators, journalists, and editors, both in print and electronic media.
Thomas Barat is a journalist, photojournalist, foreign correspondent/editor and Head of the NY Branch Office of a Hungarian-based Press Agency (WBPI) and TV Channel (Heti TV) in New York. He is the Founder Editor in Chief of CCO MAGAZINE - the Magazine of the Chief Communication Officers. He has expertise in writing, editing and publishing.
He was TV talk show host on one of Hungary's tv-channel called Budapest Television. Thomas Barat is a retired Professor of Communication and he was the Education Director of European Media and Communication Institute. He wrote 17 books in the field of Applied Communication. Mr. Barat was the President of the Ethical Committee of the Association of Hungarian Journalists and also the President of the Media Self-Regulatory Body of Hungary. Thomas Barat is a member of different professional organizations.
Among others: Society of Professional Journalists USA, Association of Hungarian Journalists, Hungarian Public Relations Association, Chartered Institute of Public Relations UK, he is the Founder and President the American Hungarian Chamber of Commerce.