The advent of generative artificial intelligence, GenAI, has changed how businesses use digital technologies. Where for many years AI was used as a predictive, analytical, and diagnostic tool, now it can produce ideas, articles, computer code, images, video, and music.
The turning point perhaps came in late 2022 with the public release of systems such as ChatGPT. These new tools allowed users to interact with complex AI models through conversational prompts. They could give the GenAI written, and more recently, spoken instructions, and the system would respond. These tools have since then become increasingly sophisticated and are now used across the corporate world and beyond.
The change happened partly because there were major developments in machine learning, a branch of computer science in which algorithms learn patterns from large datasets and can produce an output to a given prompt based on what they have learned. Central to this process is the so-called transformer model. This is a type of neural network architecture that can analyse relationships between different entries in a large volume of data. Neural networks are computational systems loosely inspired by the structure of the human brain. Transformer-based systems, including the GPT family of models, are particularly effective at generating coherent language from their training data given an appropriate prompt.
There are other approaches to GenAI. Generative adversarial networks (GANs), for instance, use two neural networks that play off each other. One creates synthetic data based on its training, and the second evaluates how real that data is based on its own training. The process goes back and forth until the output is deemed optimal and the system can no longer improve the synthetic output or make it any more real than it is.
There are various other approaches, such as variational autoencoders, which compress and simplify data and then generate variations on the themes. Diffusion models, widely used for image generation, begin with random noise and gradually transform it into structured images. More often than not, a GenAI might be using at least two of these approaches in a multimodal system that can produce text, images, and audio together.
Writing in the International Journal of Generative Artificial Intelligence in Business, researchers discuss how well all of these systems work, the value they create, and the ethics associated with GenAI. Where GenAI is augmenting one-on-one human interaction or helping make business decisions, there are issues of bias inherent in training data as well as labour disruption to consider.
As AI systems assist increasingly in analytic, writing, and creative work, knowledge workers and many other people will collaborate more and more with machines. The change is disruptive, it is likely that many jobs will become redundant. However, with automation there will be a greater need for critical thinking and ethical judgement.
Zouaghi, I. and Fosso Wamba, S. (2026) ‘Business transformation in the age of generative AI: from strategy to societal impact’, Int. J. Generative Artificial Intelligence in Business, Vol. 1, Nos. 1/2, pp.238–262.
No comments:
Post a Comment