This article serves as another reminder that generative AI should not be used for filling your official website's blog, learning business information, or for other critical tasks that do not tolerate errors.
Generative AI remains a major topic of discussion in recent months. It was the most talked-about artificial intelligence at the Collision conference in Toronto in 2023 (read more about it here), and it is a frequent subject in our blog.
However, it's important to understand that any generative AI today is very imperfect. On one hand, it does not always produce the result the user expects. On the other hand, it often presents patently false information as truth, based on facts that are not real.
If you use ChatGPT, just look at the bottom part of its interface. OpenAI, the company that develops this system, warns all its regular users for a reason: 'ChatGPT can make mistakes. Consider checking important information.'
Nevertheless, it's important to understand that it's not just ChatGPT that makes mistakes. Furthermore, it's not only OpenAI's artificial intelligence that errs. Other systems are also prone to errors. Indeed, this is why on November 15, 2023, the Cambridge Dictionary declared 'hallucinate' as the word of the year (source) — describing the process when AI presents patently false facts as truth.
In the blog, specialists emphasized that the word is taken not in its usual sense describing human sensations. For word of the year, the term 'hallucinate' associated with artificial intelligence was chosen. It denotes a situation where a neural network presents false information.
Previously, representatives of the British English dictionary Collins declared 'artificial intelligence' as the word of the year for 2023 (source). As the compilers note, the technology is rapidly developing and could become the next great technological revolution. Most specialists in tech and other industries agree with this.
We at Focus21 are confident that AI has not just become, but already is an important assistant in various fields. To say the least, even the cover for this and some other recent articles on the site was generated using ChatGPT, then refined using the powerful Topaz neural network.
But without a specialist controlling generative AI, it is only capable of creating an illusion of omnipotence and universality. Without professional oversight, any neural network today is essentially incapable of anything. We are more than confident about this (read our extended opinion on this topic here).
Our company is a space where ideas flourish and transform into reality.
