AI Is a Tool, Not a Threat; Human + AI > AI

By Troy Lowry

Is AI the future of employment or the end of it? There are many horror stories about how AI is coming for your job. While rumors swirl about AI taking over, the truth is more nuanced.1

Due to technological change, jobs from the early 1970s bear little resemblance to what we see today. Many job types from that era, like typing pools, elevator operators, and telephone switchers, have all but vanished. Back then, the Automated Teller Machine, or ATM, was a groundbreaking innovation. I recall using ATMs mainly because bank lines were usually long despite an army of tellers at the counters. These days, walk into a bank, and you'll often find just one teller and a few customers. Technology has made most everything self-service.

In 1995, applying to law school was a paper-only process. Applicants would obtain a physical application from the school, fill it out by hand, and then mail it back. Once received, the application would join various piles — those accepted, rejected, or under review. Piles, piles, piles. At that time, computer systems mainly helped track which papers were in which pile. Modifying an application was a tedious task, requiring a search through these towering stacks. Losing a few applications along the way wasn't out of the ordinary.

Now, of course, all of these processes are digital, and you can review files on your iPad while sitting on your sofa.2 This was achieved by a steady innovation in processes, facilitated by LSAC’s ACES2 and Unite products. This has not been a quick process. Although paperless processing of applications has been available in ACES2 since 2004, as recently as 2017, LSAC printed over 17 million pieces of paper (over 85,000 per law school!) in support of the admissions process.

I think artificial intelligence will gradually transform the way schools handle admissions. Some schools will adopt these changes faster than others, but within 10 years, the whole system will look completely different. It won't just be a small tweak; it'll be a game-changing shift in how schools pick their students.

AI will not, however, be running admissions offices in the next decade. Rather, it will be a set of tools, just like computers are now, with the admissions officers firmly in charge.

In a previous blog, I showed how ChatGPT can already be used to help admissions offices more quickly write communications. This is just the beginning. Schools will soon be using AI-powered bots that will allow potential applicants 24-hour access to tools that better match them to the schools that would be best for them.

I also foresee AI saving time in reviewing files by presenting summaries of the application, including specific areas that standout, to the reviewer.

One area I don’t foresee AI impacting admissions is in the decision-making process itself. Currently, many companies claim to have AI that can make better predictions about how much scholarship money people will need to come to your school.

Using AI for this process is highly problematic in several ways:

  • The nature of neural networks used in AI makes understanding how AI comes to its conclusion highly challenging. I recently read a 19 page research article that figured out how a single word was predicted.
  • AI has biases based on the data it was trained on. Because we don’t know how AI reaches its conclusions, we don’t even know what these biases are!
  • Most AI models require large amounts of data to be effective — often millions or billions of examples. Even the largest schools only get around 11,000 applications annually, with three quarters of schools receiving under 2,500 applications per year. In short, there just isn’t enough data for the AI to be able to make good predictions.
  • There’s also considerable evidence that predictive modeling using AI has considerable drawbacks even when there is enough data.

In summary, while AI will undoubtedly redefine many professional landscapes, human input remains invaluable. Technological advancements might have reduced the number of people needed for menial tasks, but they've also freed professionals to focus on more meaningful aspects of their jobs. Whether we call it "working from home" or "living at work," the key takeaway is that technology, including AI, is a tool, not a replacement for human expertise and decision-making.

  1. In my opinion, it’s unfortunate that AI is NOT coming for most of our jobs. A future in which machines do all the work in a sustainable and humane way while humans relax and pursue the arts and philosophy seems idyllic to me.
  2. Some people call this “working from home.” I call it “living at work.”