The evolution of finding data
In a recent article Seth Godin identifies that the sorting of information into directories had a big impact. He gives the examples of the phone directory, TV guide and even department stores.
Directories transform consumption
Libraries used to have index cards that helped you locate the book you were after. They were difficult to maintain and the information may have needed to be duplicated to allow searching by title or author but it provided a starting point.
We also saw this with the Internet.
Initially it was difficult to find information. Technologies like gopher and the world wide web allowed the direct navigation between information if they had a link between them. This changed even more when sites like Yahoo introduced a form of curation and effectively published a directory of links grouped by categories. We relied on the producer of the directory to curate the information in a way that made it accessible to us.
I would propose that the curation and sorting of information into directories, or lists, was just one step in terms of locating information.
The next step was searching. Yahoo was a directory of websites but it’s importance and traffic plummeted when search engines became usable enough to become mainstream. It was generally quicker and more comprehensive to search for what you needed, rather than browsing through lots of lists. By searching for a bunch of words that may appear in the information you are looking for you could locate a selection of potential results. It was up to you to pick the best result.
As commercial pressures increased the quality of the results were skewed to give higher importance to advertised information and the searches became less useful, or you had to trawl through more results before finding what you wanted.
Is it about to change again? Are we about to take the next step?
Could AI bring in the era of “asking”, where you ask for what you need and a single, hopefully correct, answer is returned?
It potentially allows you to ask in your natural language and provides you with a single answer that removes the need to work your way through a long list of results. However you are now completely at the mercy of what the AI deems to be the answer to your “ask”.
At the moment there are questions about how good it is at determining the result you are looking for, as well as questions about hallucinations and biases.
However over time this should only improve. There remains the question, will commercial, or even political, pressures in the future impact the quality of the answers it returns.
We will have to wait and see … and question the answers we get.
Links