The recent advancements in AI have already caused disturbances in certain fields of work, such as graphic design and illustration, however, predictions of its likely effects touch upon almost all knowledge-based jobs available today, with Goldman Sachs reporting 300 million jobs could be lost due to AI. Consequently, it is of no surprise that individuals are concerned for their future, especially given the rate at which AI is advancing.
Currently, the available technologies are largely Generative AI Models, which go one step further than Predictive Models and can generate new content that wasn’t in the original dataset they were trained on. Although they are still technically predicting what word or pixel should be produced next, utilising Large Language Models and complex statistical models. The next step that Open AI is working towards is the creation of Artificial General Intelligence (AGI), which is, as Sam Altman says, “anything generally smarter than humans”, however, there is much scepticism around AGI, with many arguing that a computer could never be as intelligent as a human due to the complexities of life. Yet this perspective excludes the fact that we as humans could be viewed as highly trained models, so in that sense, there is no reason why, with enough training, current models could achieve AGI (it should be noted that much of the debate around AGI is down to the definition).
The current capabilities of AI programs are already staggering, with improvements in
text-to-code capabilities, represented through GitHub Copilot, which is increasing in popularity, whilst Chat GPT can produce code too (albeit not perfectly yet). However, these text-to-code models are in their infancy and signify the mammoth industry shifts AI will cause, as these models exponentially improve over time due to more training. Furthermore, there are even text-to-video AI tools emerging such as Pika, which has outstanding capabilities and will eventually allow for high-quality films to be made through text alone.
These new technologies will undoubtedly result in job loss in many areas, however, the changes may also result in a shifting nature of jobs. As a higher quality of product can be achieved in less time using AI, yet competition in industries will still exist, so to beat competitors it is likely that, in the short term at least, instead of absorbing newfound profits, the bar gets raised, and firms end up hiring more individuals so they can provide an even better product or service to beat competitors.
Moreover, a 2022 paper by Autor et.al, assessing the “origins and content of new work” from 1940 -2018, found that 60% of workers in 2018 were employed in occupations that didn’t exist in 1940. Therefore, it can be seen why many remain optimistic about the impacts of AI, with the creation of a new range of jobs perhaps on the horizon. However, this perspective is not immune to criticism, as we have never been exposed to a technology that could eventually do all a human can, so it could be argued that these developments in AI are incomparable to any mechanical or technological shifts we have seen before.
Moreover, long-term, once these models have had further years of training, it is hard to see a world in which AI has not taken over most jobs and thus resulted in some form of UBI being implemented. This poses a huge risk in terms of income inequality, as at this point, we are in or nearing a post-labour economy, and thus it would be increasingly challenging to break out of your given level of wealth due to the lack of jobs available. Therefore, although AI could have positive societal impacts and there are currently huge opportunities to be seized, it is very much a double-edged sword, and will without question be highly disruptive to a plethora of industries in the coming years.
Comments