For over thirty years, the internet has served as a medium for sharing the collective memories of humanity. News, knowledge, creativity, research and personal experiences can be shared on the internet across all boundaries. It has supported the growth of curated digital collections: articles, reviews, photographs, web pages, tutorials and discussions that are referred to by billions of people. This great resource may now be undergoing a new phase of transformation with great challenges and fewer extraordinary opportunities.
The first obstacle this new phase of the internet may face is the rise of quickly developing generative AI. It is becoming increasingly simple to write an article, generate an image, create a video, or code a computer programme all in a matter of seconds. With the large, transformative impact of this new technology in the workplace, the ability to streamline automation has been significantly improved. While the advantages of this new technology in the workplace may be great, it has also produced a new phenomenon: the rapid introduction and increasing prevalence of AI-diluted content on the public Internet.
AI rapidly generates and posts millions of blog articles, product reviews, social media updates and even videos and images. The large quantity of this content may be mostly harmless and beneficial, but it now hurts people's ability to identify substantive content created by a human. Some researchers have even suggested that the "Dead Internet" phenomenon is contributing to the lack of genuine human interaction as the quiet is filled with automated content. This makes searches more time-consuming and annoying. However, the measurable impact is worse than that. Most modern AI systems learn from sifting through vast amounts of public data. If the AI is especially good, it is likely because its training data is especially good.
As more content is generated by AI and uploaded to the Internet, future AI systems will learn more from content generated by prior AI systems than from content created by people. This phenomenon, referred to by experts as model collapse or data cannibalism, will feed back to reduce AI outputs in the areas of creativity and language use, but will also show hallucination, or a tendency to fabricate content and errors.
This will erode trust for the average internet user. A simple search will likely yield convincing and seemingly authoritative articles that are false or misleading. AI will further the generation of convincing fake images and will fake or edit audio and video. The combination of these will deepen the deception and increase the difficulty of discerning the truth. If combined with malicious intent, this will sway public opinion, cause financial manipulation of markets and damage the integrity of AI-generated content.
While these examples will create a negative perception of AI, the opposite is true. Many fields, especially medicine, education and customer service, have learned to rely on the many positive innovations of AI. The goal has to be the responsible and ethical use of AI while maintaining the integrity of the information frameworks of the Internet that People and other non-AI systems rely on.
Governments, IT firms and international scholars are developing new strategies and innovative techniques, like secure digital asset tracking systems, responsible AI frameworks, verified digital IDs for content producers and cryptographic watermarking for AI-generated material. Instead of restricting innovation, these safeguards are meant to promote it.
The coming internet developments cannot be determined only by the development of AI. They will be determined by our collective ability to protect digital trust. The integration of AI into daily life will devalue less authentic information. In contrast, human-authored content, fact-checked journalism and verified information will be the most prized. In the future, the Internet may not be valued by its access to information, but by how trustworthy the information is. The Internet's affordability will help to build that trust. Maintaining that trust will be a primary responsibility of the AI era.
Dr Mythili Kolluru
PhD, SPP
Co-author details KV Ch Madhu Sudhana Rao
AI researcher based in the USA.
Oman Observer is now on the WhatsApp channel. Click here
