Where will AI be used?
To some extent I make a living from figuring out where and how to make use of technology, specifically computers. Here are my thoughts on where we are going in the near future with AI.
If you thought that computers are everywhere you haven’t seen anything yet, the major movement that has already started and will accelerate is the application of AI in new fields. There is much work today that can be either assisted or replaced by AI. Assisting will be the less controversial path of course but it will only lead to acceptance of AI where it performs well or identification of problems where it does not yet perform well. For anyone paid for a service it will not be up to them to decide if they are going to be replaced or not, the buyer of that service will, and that will in most cases be biased towards the quality of the service. Few will have the luxury of opting for something that is worse at a higher cost because they don’t like AI. This area of computing will by far outweigh the current area of computing, figuring out how to efficiently provide this computing power will be a problem, but not as big as one might think. For any AI to be highly cost efficient it needs to be able to do much with little resources, for any application where the human brain is more efficient or effective we will most likely continue for a while with brains.
The second area where we already use computers will need to be analyzed to figure out how it can be made better directly or indirectly with AI. There will be a high demand for humans that can do the integration work to add AI to the software that can be improved with AI. The other class of software that will not directly be made better with AI will most likely be very well defined software where repetition of simple operations and precision is dominating. These are the types of problems that we put computing at work very early, and current computers already excel. For these types of problems AI will be an indirect improvement, AI will be used to improve the design or efficiency of these systems but there will still be little to no AI in the system itself. These systems are likely to be developed in a process assisted by AI, and when the end product is satisfactory then AI will take over and optimize the system to something else. Maybe this will be similar to prototyping in high level languages for AI to then optimize into highly efficient systems. Just as I see no Assembly language developers at the office there will be fewer and fewer using performance and efficiency related languages. In a few generations we might see programming languages and runtimes that are mostly used by AI’s themselves.
There is no telling how fast this transition will be, after all, there are still humans in this process. There are major disparities in the world today, will AI reduce those? Likely not, AI requires too many fundamental conditions that need to be in place to leverage it. Leapfrogging from not taking advantage of computers today to exploiting AI for your own benefit is likely going to be the exception. However, we can hope that just as the global access to telecommunications, the availability of AI will be a net improvement for all. With the threat of “all” including bad actors of all kinds. There will likely be a new disparity, those that can benefit from AI and those who don’t. The reasonable solution to this is not to ban or avoid adoption of AI, it will have to be a solution defined by society. The last decision we should leave to AI is where and how to use it, or what we do with the benefits we can get from it.
Comments
Post a Comment