ChatGPT, the dilemma of innovation and massive layoffs

We have been enlightened by Clayton Christensen for more than 20 years on the risks of "resting on our laurels" while incumbents persevered until......resting on their laurels. So it goes.

As Clayton describes the phenomenon, the argumentative recurrence inherent in scientific development seems alien to self-acceptance due to human needs: Maslow is present and knocks down giants.

ChatGPT (Chat Generation Pre-trained Transformer) is a generative chat that is pre-trained from a method called "Transformer". Let's take a look.

Chatbots are interfaces between computers and users interested in obtaining a new product based on certain specifications. You can make the new product in the form of a text, an answer, or an explanation, or you can make it in the form of an image or illustration.

Based on the implementation of the "Transformer" method, the algorithms are developed from which they are later programmed into the computer(s) through which the user communicates, generally through text or references to other digitized objects, and for now.

Transformer is composed of (deep) learning elements that consume billions of correlated data (available on the internet) about the most significant elements provided by a user. Upon asking ChatGPT to write a poem about light, it will search and highlight in its response elements that have been successful in previous queries or with matching examples on nouns or adjectives that allow for a delicate response.

Other tools can generate images based on specific requirements. For example, DALL-E can create an elephant dancing in a Lucas film.

It is imperative to consider the true value (or threats) that these tools can create in society and, in particular, in academic and legal settings.

Many years have passed since the tools were developed, but they have certainly evolved over time. As a result of the large amount of information that these systems can ingest and process (learn), we are now experiencing an unprecedented momentum. Although these tools have not offered mechanisms that would allow anyone with a smartphone to make experiments and be surprised, consumers will decide and ponder whether the products we have are successful or not.

There is a debate about whether these systems are intelligent or not, but that debate is far from the world of applications. We shouldn't assume - for now - that neither platform is inescapable or that their advice is perfect: they will give us matching answers with their database, but they will not provide anything beyond what humans can conclude with the same database. Taking these tools as the absolute solution to understanding intelligence seems premature at the moment.

These systems will proceed through phases of personalization, superspecialization, the use of advanced and powerful mathematical tools and methods that can be implemented much faster than traditional methods, as well as the generation of more precise diagnoses based on more information.

Our industry has seen massive layoffs since the end of 2022 and even more so since the start of this year, precisely at the companies that support these technologies with multimillion-dollar investments. There are a large number of liquidated personnel who - I know - have experience and skills in developing these types of new products and companies.?

It is very likely that these new incumbents will confirm Clayton Christensen's theories once again.