Large Machine Learning Generative Models: Technological Revolution vs. Corrosive Social Impacts

The generative capabilities of LLMs are impressive. The models will continue to grow in size and applications. The contemporary technological world is in the midst of a transformation in many professional areas. It is relatively easy to see the fields affected the most by generative AI. For example, text generation will affect writers; graphics design will undergo a pervasive change, and generative music models will impact artists. 

The new AI applications will affect our perception of the world. Many negative examples of "deep-fake"  applications make us question the reality and honesty of information providers. At the same time, large models that compress enormous amounts of information provide an opportunity to accelerate scientific discovery, which currently suffers from our limited abilities to assimilate data. 

All the changes driven by new technologies are happening without understanding the long-term effects, including their implications for the economy or society. As a result, we will need guidance from visionary authors who often operate in non-scientific realms. For example, the world depicted by Frank Herbert in the Dune saga is rooted in the premise that artificial intelligence needs to be outlawed. The renowned writer sees two key reasons we are beginning to understand right now. Advanced AI can not be contained, and humanity will stagnate if intelligent machines take over most tasks. 

The challenges are visible even outside the most recent advances in machine learning. For example, careful book reading and engaging in critical thinking that comprises a large body of information have become unusual acts even in higher education. Instead, the focus is on the fast delivery of small chunks of information in a high-bandwidth format. We all know that intense focus on reading and writing helps to cultivate many positive brain developmental areas. In addition, finding the distance from many external stimuli over extended periods improves memory functions, creativity, and critical thinking. Unfortunately, these conditions are not existing for most people who multitask between work, apps, streams, interactive games, chats, and many other channels. 

 It appears we do not have a positive grasp of constructive and negative habits encouraged by new technologies. Earlier transitions from oral to written culture generated enormous benefits of democratized literacy from the printing press technology. The contemporary shift from print to high-bandwidth visual media appears questionable (as suggested by test results in schools).

What risks are ahead of us in the era dominated by automating large chunks of crucial tasks and creative endeavors? Is our culture in the process of losing its "deep literacy," as described by Adam Garfinkle in his essay on "The Erosion of Deep Literacy" in National Affairs, no. 43, 2020?

Some authors like Hugo DeGariss or Ray Kurzweil looked ahead and concluded that exponential progress would generate profound changes in society, especially when politicians get serious about using AI as a strategic tool comparable to the energy or defense sectors. New moral philosophies in favor and opposing AI will indeed become an arena of political debate.   

The architects of AI systems think of their inventions as ingenious means of bringing progress to the world. However, the rapid pace of technological advancement in AI gives new significance and urgency to many crucial questions confronting scientists, politicians, and corporate leaders about responsible applications, AI ethics, or economic impacts.  

https://nationalaffairs.com/publications/detail/the-erosion-of-deep-literacy

#LLM #GPT

https://www.linkedin.com/pulse/large-machine-learning-generative-models-revolution-vs-brzezinski