Generative AI does facilitate the work and improves productivity. It can generate reports and write the code. However, at the same time LLMs by themselves do require more human labour than what effort is saved.
Peter Cappelli, a management professor of Ivy League institute Wharton School, Pennsylvania spoke at an MIT event. He was of the opinion that LLMs create more work for people. AI is praised as a game-changing technology. There are rosy predictions about autonomous cars and tracks. They have not yet seen the light of the day. Many things are lost in the details — issues such as regulation of such vehicles, insurance liability, software issues. Besides, a truckdriver just does not stop at driving. He does a lot of other tasks. Similarly, programmers do many other things apart from programming — setting project goals, negotiating budgets etc. There are technological possibilities. However, the roll-out is slow on account of realities on the ground.
AI generates new work. Databases are to be managed. Materials need to be organized. Reports are to be made. There are issues of validity. To do all this, one has to complete many new tasks.
We have been familiar with operational AI. It is still work-in-progress. ML is still underused. There are data science issues. Data is to be analyzed. Data in silos should be integrated.
LLMs can do many tasks. However, they should avoid doing certain tasks. Letters generated by LLMs having legal implications require vetting by lawyers. Can that be a time saver then?
LLMs are expensive. They require space, electricity and manpower. It is not necessary to replace rote automation with AI.
Generative AI output requires validation of accuracy. These are to be vetted by an expert. At times, there are hallucinations and quirky responses. It is an issue of reliability.
There could be different and varied responses from LLMs. This is also a reliability issue.
People still prefer to make decisions on the basis of gut feelings or personal preferences.
In near future, most generative AI will be used to sift data and do analysis to facilitate decision-making processes.
Leave a Reply