Since the arrival of ChatGPT, generative AI has exploded, but growing disillusionment is now leading companies to doubt its real effectiveness.
Although organizations want to continue using AI, they realize that their initial expectations were too optimistic.
Suppliers and businesses expected an immediate improvement of 30 to 40% productivity. However, the results obtained do not live up to these promises. In fact, some companies have even suffered unexpected productivity losses. Additionally, AI models continue to produce “hallucinations”with unchecked errors.
Suppliers have increased their promises by aggressively promoting the benefits of AI. This competitive pressure has pushed certain players to overestimate the performance of their products. Companies, influenced by this marketing hype, made hasty decisions. They are often driven by emotions rather than rigorous evaluations.
AI implementation also generates complex governance challenges. Guaranteeing the confidentiality, security and explainability of models requires careful work. Many companies had not anticipated these additional efforts. In addition, the processing of unstructured data, combined with AI modelsrequires solid infrastructure and time.
Slower adoption but no abandonment
Although disillusionment is slowing adoption, companies are not abandoning generative AI. They become more demanding and rigorous in their tests and evaluations. They are now seeking to better understand the return on investment before making a full commitment. Adoption continues, but it now relies on in-depth analytics rather than initial enthusiasm.
To overcome this disillusionment, I think suppliers need to adapt their business strategy. In my eyes, they should sell their products by highlighting their real capabilities and clearly demonstrating their value. Integrating security and privacy into their solutions would also make implementation easier for customers. Additionally, data challenges can be turned into opportunities. For example, automation tools could optimize the labeling and classification of unstructured data.
Share the article:
Facebook
LinkedIn
Our blog is powered by readers. When you purchase through links on our site, we may earn an affiliate commission.