Wyatt Bevis
When we first covered the topic of metadata I was interested in where it originated and how it was collected. Then, after a bit of research, I now had the context of the early development of metadata and I started to wonder how the technology would influence metadata collection in the future. In my blogpost, I will attempt to foresee problems in the field of metadata an propose solutions to them.
Before discussing the problems with metadata, I will provide some background information to help provide context. The term “metadata” first appeared around the 1960s in the realm of computer science. However, the idea of storing information to organize data stems from libraries dating back to the ancient world. Today, metadata can be found basically in any place where data exists, for example, public records, image files, emails, digital libraries, and websites all utilize metadata.
With the development of AI, it seems most metadata collection is shifting from manual to automated entry. In a lot of ways, this is great. AI will be able to ensure that metadata stays updated with its information. Moreover, individual biases can lead to miscategorization of metadata especially when cataloging controversial events. For instance, if a more biased person is creating metadata for a divisive protest they might instead attach the word riot to the event. Which would in turn make the event more difficult to find. Conversely, like most technological improvements, automation of metadata collection will lead to a loss of jobs for some people. It seems to me that this is an unfortunate cost of technological progress that there is no plausible way of avoiding without foregoing all the benefits of automation. AI automation has been known to produce mistakes occasionally. AI often misrepresents metadata for images when it fails to identify something a human would always recognize. This could be with anything from an image with bad quality or something subjective like culturally significant things.
Additionally, AI relies on machine translations when switching metadata between languages. This oftentimes becomes inaccurate when translate words with multiple meanings. Recent experimenting with AI has uncovered a solution to this problem. Instead of attempting to directly translate source material, success has been found in “linking relevant concepts cross languages”(Qin, 2014).
Did you know that when you enter a companies website you are granting them access to information including your location, browser history, personal details, and even your social media accounts? These companies even attempt to predict your activity. It is overtly evident that privacy is often overlooked in the collection of metadata. But what can we do about this? Well, first of all the extent to which data is collected about individuals needs to become common knowledge. This could be accomplished simply by telling other people. We would also all benefit from disabling cookies from our browsers. This would eliminate much of the problem for people who value their privacy over online convenience. Although, turning off your cookies still is not enough for some sites. Meta, for example, will continue to harvest your personal data as long as you use there site, and even if you stop they will keep a record from your previous activity. This is why as a last resort we must advocate for a policy change on the government level.
Metadata, or information on information, is not without its problems. Topics such as language barriers, automation, and privacy deserve our attention. In conclusion, metadata effects our lives in so many ways that it is almost impossible to track all of them.
Reflection
The LLM I use gave me great pointers on how I could have improved my writing. First, It recommended I include a hook that could engage a broader audience. It also pointed out when I was vague or abrupt with a point. Finally, the AI showed me that I flipped from being formal to informal. I noticed that the AI had a hard time understanding who the audience for my writing is. From the bit of research I did, I learned a lot about AI “thinks” or operates.
Leave a Reply