Can You Use Generative AI for College Searches?

Over the weekend, I discussed my nephew’s interest in colleges with my sister-in-law. He is interested in engineering programs, and my sister-in-law said he is interested in engineering because he enjoys making things around the house and at school. Our discussion turned to using generative AI for college searches.

Marco (my nephew) is the second child, and his older brother is a rising junior in college. His mother has been busy investigating potential universities. She created a spreadsheet of universities with facts pertinent to the decision process.

Marco’s older brother was interested in biology programs, so most of the universities on the list she built were new to his parents. Besides engineering programs, Marco’s parents are interested in first-to-second-year retention and four-year graduation rates. They are also interested in the college’s percentage of Hispanic students.

I asked my sister-in-law why she was interested in four-year graduation rates. She explained that many of the engineering schools on her list are at public universities. She is also concerned that a reputed lack of course availability at public institutions could delay her son’s on-time graduation.

There are 18 universities on Marco’s list. I appended the list below and four pertinent data points for each. My sister-in-law said that she retrieved the data from The College Navigator website. I randomly checked a few data points.

I decided not to consult a source like U.S. News & World Report’s Ranking of Engineering Schools in order to verify or add schools to their list. Instead, I decided to use two generative AI apps, Claude AI (3.5 Sonnet) and ChatGPT-4o, to see if some colleges or universities should be added. I used the same prompt for both apps. The prompt and responses are appropriately labeled.

Prompt

My nephew is a rising senior in high school. His parents are interested in having him attend a college that will embrace Hispanic students. He is interested in engineering, specifically because he enjoys making things with his hands. Can you recommend universities with engineering programs that have a decent percentage of Hispanic students?

Claude AI (3.5 Sonnet)

I understand you’re looking for university recommendations for your nephew, considering his background and interests. Here are some suggestions for engineering programs with a significant Hispanic student population:

1. University of Texas at El Paso (UTEP): Known for its high Hispanic enrollment and strong engineering programs.
2. Florida International University (FIU): Has a large Hispanic student body and offers various engineering majors.
3. California State Polytechnic University, Pomona: Known for its hands-on approach to engineering and diverse student population.
4. University of California, Riverside: Has a good engineering program and a significant Hispanic student population.
5. University of New Mexico: Offers strong engineering programs and has a high percentage of Hispanic students.
6. Texas A&M University: While not predominantly Hispanic, it has a sizable Hispanic population and excellent engineering programs.
7. Arizona State University: Has a growing Hispanic student population and well-regarded engineering programs.
8. San Diego State University: Offers various engineering majors and has a diverse student body.

These universities have engineering programs that might align with your nephew’s SAT scores, though admission requirements can vary. It’s important to note that many universities consider factors beyond test scores, including extracurricular activities, essays, and overall academic performance.

I’d recommend researching these schools further to see which ones best fit your nephew’s specific interests and goals. Consider factors like location, campus culture, specific engineering specialties offered, and opportunities for hands-on learning.

Would you like more information on any of these universities or advice on other factors to consider in the college search process?

ChatGPT-4o

There are several universities with strong engineering programs and a decent percentage of Hispanic students that might be a good fit for your nephew. Here are a few recommendations:

1. University of Florida (UF)
   • Engineering Program: UF has a highly ranked College of Engineering.
   • Hispanic Population: UF has a significant Hispanic student population, and it is recognized as a Hispanic-Serving Institution (HSI).
2. University of Texas at Austin (UT Austin)
   • Engineering Program: UT Austin’s Cockrell School of Engineering is well-regarded nationally.
   • Hispanic Population: UT Austin has a large and diverse student body with a substantial percentage of Hispanic students.
3. Arizona State University (ASU)
   • Engineering Program: ASU’s Ira A. Fulton Schools of Engineering offers a variety of engineering disciplines.
   • Hispanic Population: ASU has a diverse student body with a notable Hispanic population and is an HSI.
4. California State University, Long Beach (CSULB)
   • Engineering Program: CSULB offers strong engineering programs and hands-on opportunities.
   • Hispanic Population: CSULB is an HSI and has a significant percentage of Hispanic students.
5. Texas A&M University
   • Engineering Program: Texas A&M’s College of Engineering is one of the largest and most respected in the country.
   • Hispanic Population: Texas A&M has a considerable Hispanic student population and various support programs for minority students.
6. University of California, Irvine (UC Irvine)
   • Engineering Program: UC Irvine’s Henry Samueli School of Engineering is highly ranked.
   • Hispanic Population: UC Irvine is also recognized as an HSI and has a strong support system for Hispanic students.
7. Florida International University (FIU)
   • Engineering Program: FIU’s College of Engineering and Computing offers a wide range of engineering programs.
   • Hispanic Population: FIU has a very high percentage of Hispanic students and is an HSI.
8. University of New Mexico (UNM)
   • Engineering Program: UNM’s School of Engineering is well-regarded.
   • Hispanic Population: UNM has a significant Hispanic student population and is an HSI.

These universities not only offer strong engineering programs but also have supportive environments for Hispanic students. It might be beneficial for your nephew to visit these campuses, if possible, to get a feel for the environment and resources available.

Analysis

Claude 3.5 Sonnet provided me with a list of eight universities. Only one of these, Texas A&M University, was on my sister-in-law’s list. I accessed College Navigator to find the information that my sister-in-law had included on Marco’s list. Adding these schools to the spreadsheet provided a helpful overview. It’s too bad Claude didn’t provide that data or its sources verifying reputation and Hispanic student percentage.

Six of the remaining seven universities on the Claude-generated list appeared to focus on the percentage of Hispanic students. They ranged from a high of 88% at the University of Texas at El Paso to a low of 26% at Arizona State University.

Claude also included a decent wrap-up statement recommending researching the schools to see which one’s meet my nephew’s specific interests and goals. It also recommended considering factors like location, campus culture, specific engineering specialties offered, and opportunities for hands-on learning. Many of those data points are in my sister-in-law’s spreadsheet.

I did not specify a specific number of universities, so I assume it was purely a coincidence that ChatGPT-4o provided me with a list of eight universities. Again, only Texas A&M showed up on both the list my sister-in-law created and those suggested by the AI responses. Four universities on the Chat GPT4o list were on Claude 3.5’s list. Again, I accessed data from College Navigator to add and compare with Marco’s list.

ChatGPT-4o recommended visiting campuses, if possible, to get a feel for the environment and resources available.

Unfortunately, neither Claude 3.5 Sonnet nor ChatGPT-4o gave me rankings of the engineering programs at any of the recommended schools. Interestingly, my sister-in-law did not include rankings in her dataset of items for each university’s profile. However, her choices of retention and grad rates are the highest weightings in U.S. News’ undergraduate rankings.

I created a table with all the recommended schools to provide more information and perhaps create a gap between schools. I then added a code for the original source (Marco, Claude, or ChatGPT-4), and the 2024 U.S. News & World Report Best Undergraduate Rankings. I kept the universities in the original order but could easily sort by the percentage of Hispanic students, grad rate, U.S. News rating, etc.

I could have used a rankings source other than U.S. News. However, its list of schools is longer than the Burning Glass Institute/Wall Street Journal analysis that ranks the Top Colleges for High Paying Jobs in Engineering, Finance, etc., I reviewed a year ago.

U.S. News rated 208 undergraduate engineering programs. All but three of the 29 universities in the table were rated. After looking at the rankings, I will recommend that my sister-in-law and Marco consider some schools with higher rankings than others on this list. They can look up the percentage of Hispanic students to see if it meets their minimum.

They should also look at the Burning Glass/WSJ analysis of the Top 20 Public Colleges for High Paying Jobs in Engineering. Several of the universities on their original list are listed there. If the return on investment is important to them, this may be the best list to review. A table with the top 10 is below.

A Few Final Thoughts

From the first days of ChatGPT, I’ve used AI as a tool to supplement my work. I’ve seen improvements in the output from all the generative AI tools I use. If I used it to write research papers, I would give it a grade of 70 because it’s not as well-researched as a paper I would write.

In this exercise, I used Claude and ChatGPT to supplement a list created by my sister-in-law. Surprisingly, each tool must have prioritized the percentage of Hispanic students in their search criteria. As a result, only one school overlapped all three: Texas A&M.

I didn’t include school rankings in the content requested in the prompt. Only two schools recommended by the two AI apps were not ranked. Only one on Marco’s list was unranked, but I have a feeling my sister-in-law queried a few of her friends for recommendations.

The good news is that Marco attends a very good independent school and is scheduled to meet with a college counselor over the next few weeks. I’m sure he and his mother will receive excellent advice, including the relative SATs that these universities consider. They’ll also learn more about the success of previous students from the school who have matriculated there.

I have a few friends who have advanced their prompt skills so that they can instruct the AI to locate specific data related to their query. I am not sure the data I looked up on College Navigator and U.S. News could be accessed randomly. I presume it could be loaded as a table or source for the analysis.

The 11 additional schools recommended by ChatGPT-4o and Claude 3.5 Sonnet had a percentage of Hispanic students at 24% or greater (UT El Paso is 88%). Since data sources are seldom shared in the output from these two tools, validating their selection was something I could do more easily than the average person.

While the recommended lists with no listed percentages of Hispanic students support my suggested grade of 70, they also support several studies I’ve read. These studies show that students in the bottom half of a class using a Generative AI tool increase their performance at a higher rate than students in the top half of the class using the same AI tool.

If I had started this task from scratch, I would have likely circled universities from the U.S. News list in the areas my nephew prefers (Southeast and Southwest U.S.). I would send the list to my nephew’s college counselor. They could then find out which universities they thought were possible for admission, which were a stretch, and which were red flags.

After the list was prioritized, I would research the percentage of Hispanic students using College Navigator and determine which schools to visit. Generative AI would not have been used because I can quickly find the most relevant data using traditional search tools. For others, the examples produced by GenAI in my prompts may be a very reasonable start.