I tend to agree with those people who say that AI tools such as ChatGPT and Gemini aren't yet ready to replace skilled developers. By the term AI, I do of course mean Machine Learning Models. And I'm not just restricting this to software developers. AI tools aren't there yet to replace programmers, graphics designers, accountants and voiceover artists. It can take 10 to 20 years of hands on work experience to become skilled at these careers. They are handy tools for knocking out sample code and procedural suggestions. You can certainly learn new things from their output.
However, as AI matures it will display one big advantage over a skilled human workforce.
Although there are skilled programmers, skilled graphics designers, skilled accountants, skilled project managers and so on, each one works in silo. That silo is dedicated to a particular skillset. For this workforce to jump from silo to silo is much harder. There is a huge barrier between being a skilled accountant and a skilled computer programmer. AI sees no barrier between these disciplines.
To AI tools, each silo is simply another large language model. AI will be able to pull in knowledge from the programmer silo, the accountant silo, the marketing silo, the management silo and thread them all into a single output. And that output will be at the highest level of knowledge it has of each silo.
During my IT career I worked on a variety of projects in different industries. They included investment finance, insurance, mortgages, healthcare, telecoms, book publishing, engineering and others. I applied my technical software knowledge to each project as best I could. But entering each new industry was a steep learning curve. There is nothing common between the 5,000 terms used in investment banking to the 5,000 terms used in telecoms or the 5,000 terms used in supplying mental health data. And the working procedures of a bank are very different to the working procedures of a hospital. These are silos of knowledge and data that have little overlap. Just as well for me they all relied upon SQL.
The AI tools we see today are what I would call external tools. They exist outside of a specific industry. The graphical design industry may currently be the only exception to that rule. In the future, AI tools will become more industry specific. One will be tuned to banking, another tuned to healthcare, another tuned to publishing. But in each one, the AI tool will be able to easily cross those discipline barriers.
An AI tool that understands computer programming will also be able to understand accountancy, sales, marketing, project management, supply chains, pricing models, and anything else a large language data model can be built for. It will be able to understand the latest regulations in each industry. Do you understand all of the regulations in your industry. I bet you don't.
So it is this ability to combine knowledge silos that is going to be AI's greatest strength.
Up until now, the skilled workforce has engineered the best answer to a question. In the future, these same staff will be engineering the best question for an AI tool to produce the answer.
And a last thought. AI tools will be used in HR to select the perfect candidates for employment. Now that's scary.
No comments:
Post a Comment