Are We in an AI Summer or is Winter Coming?

Ok, I know. I can't be talking about an artificial intelligence (AI) winter in the era of the massive success with chat-GPT! Trust me, I get it. But if you ask those who lived in the early 20th century, they've seen this drama with AI before - an incredible rise in hype only for hopes, public attention and funding to dry up once the hype did not meet expectations. I am a contrarian by nature, so may be a little bit pessimistic with my views but I thing a new AI winter is coming. Before I tell you why I think so however, I'll like to define what I mean by an AI summer and an AI winter

AI Summer

This happens when AI hype is rapidly increasing, with funding, public perception and interest also increasing. If you think about the recent success of chat-GPT, you'll realize that we clearly in an AI summer. Everyone is talking about the impressive feat and possibilities of artificial general intelligence (AGI) like chat-GPT and funding is flowing into OpenAI (creators of chat-GPT) like they never could have imagined when they initially started as a non-profit. In fact chat-GPT has been such a major success that in its short life span, it's completely upending humanity as we know it.

Schools are adapting their curricula and how they teach and access students in the era of chat-GPT since students can use chat-GPT to write whole essays! In fact, chat-GPT has recently passed medical and bar exams! What does all this mean for the future of educating and testing in general? Not all of this however is bad. I for one believe that our educational system is long due for an overhaul and if chat-GPT catalyses such needed change then so be it.

Software Engineers and programmers use chat-GPT to be more efficient at their jobs, so much so that there has been talk that chat-GPT will replace programmers and software engineers. I think the truth about this scenario is more complex. What chat-GPT will do is replace less skilled programmers and software engineers while upgrading the skills or medium to high skilled programmers and software engineers. The super star programmers and software engineer - think of them as the Lebron James of coding - will be largely unaffected. There might be a small uptick in their efficiency but these superstars are good enough, for chat-GPT to make a significant difference in their performance.

Talks of industry disruption are classic consequences of any technology summer and disrupting education and even the jobs that created chat-GPT itself (programming and software engineering) are fantastic examples and use cases. However, every hype lives in a cycle, with the loudest beat heard at the peak of such cycle, clearly begging the question - is this AGI hype/cycle about to go into its downturn? Are we about to enter an AI Winter? I definitely think so.

AI Winter

Technology winters happens when the hype does not meet expectations and society looses faith in the given technology. Generally technology winters happens because the given technology, driven by its initial success, then get overhyped. This overhype, in my opinion, is what we are seeing with chat-GPT. Why do I think so? Because of the general nature of Large Language Models (LLMs) that forms the foundation of chat-GPT. To succeed, such models need to be fed a ginormous amount of data. The challenge is such data have complex societal issues associated with them like privacy, ownership and intellectual property. Generally, industries have solved these kinds of problems legally by settling in and out of court, but such situations were highly localized and hence manageable. However, with chat-GPT, its data issues will be massively distributed across nations and continents. For example, Italy just banned the use of chat-GPT until some of its data issues are resolved. It doesn't take too much of a mind stretch to realize such legal and national battles will not stop at Italy.

Finally, it is clear to almost everyone watching this space that chat-GPT and AGI in general will have to be regulated by governments. The challenge is different governments will regulate and legislate AGI is different ways. However, the answer to a chat-GPT query should and must not be dependent on the country in which that query was made! How chat-GPT and its creators OpenAI will solve this problem is still open for debate but one thing is for sure. If they don't or can't solve it, we may be in for a very long AI winter.

What do you think?

Yours in learning,

AI+Finance