Please note that the opinions below are those of my own and do not reflect those of any organizations I may be affiliated with.
While discussions around generative artificial intelligence (GenAI) have evolved around speed, those discussions are increasingly surrounding the level of detail paid to the output of GenAI and who is responsible for it. Indeed, increasingly as GenAI is integrated into existing human-centric processes, what is the responsibility of GenAI and what is the responsibility of the human in the loop will be a major discussion point.
Many would argue that society is already debating the finer points of GenAI content and not in a good way. The moniker “AI slop” has come to label the low-quality content that the general public has unfortunately started to associate with AI content.
If one dives deep into the term “AI slop” what does it mean and what does it portend? According to Wikipedia, “AI slop” is digital content made with generative artificial intelligence that is perceived as lacking in effort, quality, or meaning. There is no doubt that the first iterations of material being produced by GenAI fits the definition. Whether it is written prose that makes no sense to the average reader, video images with disappearing limbs, or video content that is devoid of the emotional nuance of traditional filmmaking, everyone has seen some form of “AI slop”. The question though is who is at fault.
While many individuals would solely blame the technology, the reality is that humans need to accept some responsibility as well. Indeed, this is a pattern that has occurred with the introduction of every technology that has been introduced.
From the telephone to the television, every modern convenience that is part of daily life in society today, has been blamed for some societal upheaval that has occurred with said technology’s introduction. This anthropomorphization of technology has been a trait of human society since the dawn of time.
The reality is that the anthropomorphization of technology overshadows the fact individuals and society in general are responsible for how beneficial or detrimental a technology is used. Even in the case of GenAI.
While there is talk about the AI singularity, at present time, the advancements that have been made concerning GenAI puts humanity potentially closer to the AI singularity but no where near it. At present, it is up to humanity to separate the long-term vision from the practical realities of implementation.
What does this mean for the human-GenAI interaction? It means that humanity is still in the driver’s seat and decision making as well as the details behind said decision making are still within the control of individuals not GenAI.
While there are many who would argue that GenAI is already at a point where it can make its own decisions and is beyond the control of individuals, the reality is far different. There is no doubt that today’s iteration of GenAI is capable of doing much more than its predecessors independently.
However, at the end of the day, GenAI is still at the whim of human decision makers. Indeed, while humanity worries about the coming of Skynet, the reality is that it is still up to individuals and organizations to determine how GenAI is used.
Indeed, the rise of “AI slop” isn’t due to the prevalence of GenAI but it is due to the lack of established norms and attitudes concerning GenAI by individuals and organizations. In other words, it is due to society still determining what is acceptable GenAI usage and what is not.
GenAI is forcing a discussion concerning what is acceptable usage and what is not whether society is ready for it. Much like social media, tobacco, and driving, society is still trying to develop acceptable standards concerning what will be acceptable use and what will not.
How does this play into the fact that individuals and organizations must accept responsibility concerning the need to review GenAI output in detail? In many respects, much like other technologies and inventions before it, society is still determining what will be accepted when it comes to GenAI output, particularly when it comes to consumption by the general public. However, guidelines are forming as GenAI output is being pushed out at an incredible pace.
The advantage concerning GenAI output is twofold. First, is the ability to create and distribute output without having to convince and manage multiple stakeholders. Second, as many have stated, it is the speed at which it can generate output. Both have pitfalls that need to be addressed but it is the speed that is of most critical when it comes to detail management.
The reality is that just like with human-generated content, GenAI content doesn’t get it right on the first pass. One has to only look at geniuses such as Leonardo da Vinci or Albert Einstein to see how relevant this message is.
Their breakthrough outputs were accomplished through multiple iterations of trial and error. These iterations allowed them to improve and refine their ideas through detailed mental exercises and experimentation. Without these multiple iterations, the breakthroughs that have been made would not have been achieved.
What relevance does this have concerning GenAI outputs? The reality is that details matter whether the output is created by GenAI created or not. Unfortunately, the current generation of creators have forgotten that in favor of speed.
The old saying that “details matter” is more important than ever. Indeed, as humans are still making critical decisions, particularly when it comes to strategic ones, humans need to be convinced. To convince humans, it is necessary to “sweat the details”.
The reality is that while it is incredibly easy to ship first draft content thanks to GenAI, it is still incredibly difficult to convince people of an idea or a concept even with GenAI. Indeed, it is sometimes harder to now with GenAI as the mind share is shifting from an open stance to a more skeptical one.
As such, individuals and organizations are determining what is in their best interests when it comes to GenAI content. One of the biggest factors that they have discovered is that while over the short term it may be acceptable to ship GenAI content fast, the majority of individuals and organizations are realizing that shipping publicly slower while iterating internally faster maybe the best approach.
These individuals and organizations are realizing that sweating the details and making sure that narratives and ideas are fully baked is the best way to build long-term growth and trust with a skeptical public. While there is no doubt that fast content shipping may mean faster and higher initial consumption numbers, the reality is that such an approach isn’t sustainable or profitable in the long-term.
It isn’t that GenAI doesn’t have value. It allows more preliminary ideas and concepts to be tested faster thus allowing for better final ideas and concepts to be released publicly. If there is sufficient detail paid by humans as to the quality and coherence of the final output.