How generative AI changed the Internet in just one year? The end of the Internet as we know it?

• Generative AI has affected the digital world and sparked mild “paranoia.”
• Misinformation and computer hallucinations have also gained more space
• With the advent of artificial intelligence, dilemmas of ethics and copyright arise frequently

It has been more than a year since the Internet was filled with surreal images generated by artificial intelligence, fake state representatives promoting cryptocurrency companies or fake videos from all over the world. Although it may not seem like much to us, this phenomenon in the form of ChatGPT, Midjourney, and other models has changed the face of the Internet forever. He was November 2022, when OpenAI debuts ChatGPT in public and it becomes one of the fastest growing apps ever. It also raised many questions regarding ethics and a slight skepticism towards similar technologies began to prevail among the general public. Facebook pages were filled with unrealistic created images, which unfortunately older users often believe, and misinformation gained more space than ever before.

Artificial intelligence has dominated us for several years, whether in self-driving cars, in chess programs with predictions from 1997, in predicting the weather, in automating the production processes of car manufacturers, or simply in sending relevant results in Google searches. However, the concept of AI itself has only become popular in the last year, thanks to the advent of generative models and the possibility of trying it out for free as an individual, in the comfort of your own home, with real, up-to-the-minute results. . AI itself has been at our side for a long time, without us particularly feeling its presence.

The term hallucination has acquired a new meaning

Even computers can hallucinate. This is one of the things that 2023 has brought to our lives. AI hallucinations are the so-called generation confident of an answer that has no basis in reality. These are the answers that don't have enough data collected during training, so she just makes them up and gives the feeling that she believes them herself and that they are correct. Even Google itself once got burned when it shared in one of its ads promotional materials created by Bard, which was not true, and of course users did not leave a dry topic about this in the comments.

Bard is an experimental AI-powered chat service developed by LaMDA. Developed using our amazing language models and based on insights from the web, it serves as a launching pad for the curious and can help simplify complex topics → pic.twitter.com/JecHXVmt8l

Google Maps will lose practical functionality. Especially drivers will pay

-Google Google) February 6, 2023

Deepfakes – fake photos and videos and growing concern about their authenticity

The skeptical behavior of not trusting every photo or photo of our loved ones or celebrities has been awakened in us by Instagram filters in recent years, but generative AI is taking it all to a new level. Deep fake or not Hard-to-identify edited media has flooded the InternetFacebook groups and maybe your favorite websites too. Although it has opened the doors to the imagination of many users, it has also allowed the generation of inappropriate content, which often serves nefarious purposes or targeted violations of moral principles. One of the most notable instances of this case is the misuse of the identity of actress Scarlett Johansson, whose voice was used to promote artificial intelligence without her knowledge. A similar situation appeared in the series Black Mirror, where Salma Hayek was abused by the live streaming platform.

Security concerns about selection of training data

These language models are trained on content from almost all over the Internet. This includes everything you can imagine – from blog articles and social media posts to Wikipedia and even pirated websites. then A large amount of information is often very dangerousOnline anonymity is a powerful magic, and it is not always possible to filter or block inappropriate comments appropriately. Even with this data that LLMs (language models) work with, this information often violates privacy and copyright protections. A very disturbing case came to light recently when it was discovered by the Stanford University Internet Observatory More than 3,200 suspicious images in the databaseWhere, for example, Stable Diffusion is trained.

The study reveals the presence of CSAM in a popular AI training dataset

LAION-5B contains 1,008 verifiable instances of illegal images of children, and likely many more, researchers say. A huge public dataset that served as training data for popular AI image creators, including Stable pic.twitter.com/d9ILut5VHc

-John Marcellus (@safespace4space) December 20, 2023

Accelerate and automate business processes with generated content

Although generative artificial intelligence has already reached a high level, despite great efforts, it is often compared to a high school student who has not studied much, but is able to understand and produce content, and is prone to major errors, errors, or even more general It is necessary. He has improved a lot in one year, He can help with routine operations, come up with creative ideas, and is an ordinary, pleasant employee. The most notable case was, for example, that of articles on the Gizmodo portal, where the company began publishing content generated by artificial intelligence without subsequent production or monitoring, thus unknowingly presenting false content.

Artificial intelligence has opened a new field for business

With the advent of artificial intelligence, new business circles and opportunities have also opened up. Generators allowed new entrepreneurs to save on writing and graphic work, and even allowed for the sale of digital products. In today's world, you can make passive income online with relative ease, in the form of selling posters, e-books or even digitally generated courses, all generated entirely for you by an AI model. Humans seize the potential before AI swallows up copyright and regulation. Internet today we can With a great exaggeration compared to the post-revolution period, When a certain kind of illegality still exists it gives people the freedom to experiment and new ways of earning.

Android Auto has received a new design feature. Will you leave him?