There are limits to current automation. Would you like to completely automate your grocery deliveries to your house? During this COVID-19 pandemic, it would be more than a convenience. It may save lives from yours, to the grocery cashier, stock person, delivery truck driver, all the way to the farm workers. But, to do so would require an algorithm to understand what you’d like to stock your fridge with (before you do), a robot to select and pack items from a warehouse (including distinguishing fresh from rotten fruit, or an apple from an orange), and an autonomous truck to drive your order to your home. This can be extended to robots at the farms reducing the risks associated with a catastrophic reduction of food pipeline workforces. None of that can be automated without AI. Now, extend that to all the other problems where intelligent robots could have helped mitigate risks from losing essential workers, or from putting them at personal risk: First responders, health care, mass transit, and social services workers.
AI is also automation, but for more difficult problems. These are problems that exist in dynamic, often semi-structured or unstructured environments. To pick an apple, for example, a farming robot would need to find the apple trees, then distinguish between a ripe apple vs an unripened one vs a leaf or a twig. If all apples were exactly alike, and located in the exact same spot of every apple tree, year after year, this would be easily automated. But, the world doesn’t present itself that easily for us. The “automation” techniques used to handle these types of variations are labeled as “artificial intelligence”.
It may be more useful for you to think of automation as a spectrum of technologies/techniques. On one side of that spectrum are simple scripts and programs that have unburdened much of the highly structured rote work that a sufficiently lazy engineer would have automated (Think of using redstone in Minecraft to farm pumpkins). On the other side of that spectrum are problems that aren’t well structured, dynamic or constantly changing, or require advanced planning (Think of managing activities for kids high on sugar at a daycare center). When automation itself needs to be automated, then you’re headed towards the AI side of that spectrum.