Will AI Agents take away our jobs?

Day 363 / 366

The conclusion in the field of AI this year has been that we have reached the level of intelligence we could achieve with just Large Language Models. All big AI companies are now shifting towards utilising the abilities of existing models to create AI Agents.

What is an agent?

While LLMs are in the end just text-generation models. If you create a framework in which you give these LLMs tools that they can run and ability for them to make follow-up calls to themselves, you get yourself an AI agent.

One use of agents could be gathering information from the web. If we give the agent the ability to open web pages, we can then give it a task say to grab the current score of a match. It will be able to search for live matches on Google, look at the results and use Gen AI to return the scores.

I can definitely see AI agents being useful. In fact, I regularly use one as a coding assistant. It allows me to get trivial tasks, like writing unit tests quickly. The biggest advantage of this as compared to something like co-pilot is because the agent has the ability to run code and look at the output. And then based on the output, it can make changes to the code and keep on iterating till the task is done.

I think 2025 is going to be the year for AI agents.