Large Language Models

Diving into Artificial Intelligence

Artificial intelligence is all the rave these days. All companies had to do during this latest technology hype cycle was mention that it was focusing on artificial intelligence and investment capital flocked to them. Investors seemed to be focused on jumping in before figuring out what that investment meant. Very few investors in the space seem worried about what income these companies are actually generating from their massive investments in the space. The poster child of this whole movement is Open AI and their product Chat GPT. Its core product is artificial intelligence, and the product is based on a large language model. It creates tokens through its responses. Taking the information readily available created through the tokens and forms the information into a response. At its core it creates laziness, and in reality the response has been trained by the company that owns the large language model and the users interacting with it. For me it seems useful. It also seems like it is a way for all of humanity to disconnect with itself. Time will tell but I think it is safe to say that the technology is going to advance. People may need to embrace it or they risk becoming detached from development. I don't think anyone knows which way it will truly go, but there is a chance that humanity is on the edge of vast change or some sort of depression and reversion back to the mean.

It has the promise of taking away swaths of jobs in almost every industry. You have had CEO's come out and speculate that artificial intelligence will soon become more capable than its human counterpart. When this happens it will create a world in which all control will be held at the top of the pyramid, and everyday people will receive a universal income which could further diminish the purpose for our population. Will it ever get there. It seems very plausible that this could happen in some form, given the push to adopt and control the technology. A complete opposite narrative would be that the technology is being oversold, and that just maybe the capability will be capped by limitations within the technology. For instance these large language models may at some point corrupt themselves because the information used to train the models is not founded in credible information. If the corrupted learning compounds on itself the response may lose relevance to the prompt being entered. In my humble opinion I think we could be a decade away from a lot of the advancements being talked about. I could be very wrong. Either way it goes I feel the need to at least understand the technology that stands to take over all production.

Inevitably I believe the technology is here to stay. The largest countries in the world and the largest companies in the world have made a huge bet on the technology. It seems to big to fail. With that lays a great deal of risk. A risk that an overbuilding of the infrastructure needed will be a huge drag on company earnings which has the potential to turn financial markets on their head. This in turn could lead to a squeezing of the wealth of the top 10% of income earners along with the companies they are invested in. Which in turn has the potential to turn several asset classes into a bubble ready to pop. It is a the conundrum of our generation. Hope it succeeds and displaces huge pockets of the economy in favor of the wealthy, or hope that it lacks adoption and leads to economic turmoil unlike we have seen in generations.