Between 2004 and 2025, Google searches about skills, knowledge, leadership, AI, jobs, and education have moved from loosely related concerns into a more unified conversation about AI anxiety and career impact.
A new report by National University tracks eight search topics—knowledge gap, skill gap, leadership gap, Gen AI, ChatGPT, can’t find a job, job market, and getting a degree—to understand how people interpret employability and career risk, and how the arrival of generative AI in late 2022 appears to have reshaped that decision-making landscape.
Key Takeaways
- Worries travel in packs. Searches about skill gaps, knowledge gaps, leadership gaps, AI, and not being able to find a job tend to spike and dip at the same time. They’re all part of the same anxiety.
- ChatGPT changed the conversation. Before late 2022, job market searches were unrelated to skills anxiety. AI made these worries feel like one big problem.
- “Getting a degree” was the most-searched topic in this dataset for over 20 years. Education remains the go-to option when people feel uncertain about their futures.
How Do the Topics Move Together?
Across 2004–2025, the eight search topics form one big conversation about readiness for work, with education and capability gaps at the center. Getting a degree is consistently the top signal in the data: it is the highest‑interest query across the full period and finishes at 100 relative interest at the end of 2025.
Across the full 2004–2025 period:
- The gap terms move as a tight cluster. Over the full period, knowledge gap and skill gap have a correlation above 0.75, and leadership gap is also strongly aligned with them (correlations above 0.60 with both). When people search one of these, they are usually searching the others more as well.
- Getting a degree is strongly correlated with all three gap terms, with correlations in the 0.6–0.8 range depending on the pair. This suggests that when people feel gaps in knowledge, skills, or leadership, they are also actively considering further education as a response.
- Once they appear in the data, Gen AI and ChatGPT show positive correlations with the gap and degree series (often above 0.5 where there is enough overlap), joining the same conceptual cluster rather than forming a separate, isolated trend.
- “Can’t find a job” search term has a solid positive relationship with the gap and AI series (correlations mostly in the 0.5–0.7 band), indicating that job-search anxiety is bound up with perceived skill and knowledge deficits and with AI.
- “Job market” search term is related, but not as tightly tied to the “gap + AI + education” cluster as the others in the full-period view, acting more like a broad macro sentiment indicator.
Taken together, the correlation structure and normalized time-series patterns point to a large, shared attention cycle: skills, leadership, knowledge, degrees, AI, and job‑finding difficulty tend to move together.
The AI Era Shift
To understand how the mainstreaming of generative AI reshaped these relationships, we compare correlations before and after November 2022. In the post‑AI era, all of these search topics snap into a single, tightly connected system centered on generative AI. As Gen AI and ChatGPT surge after late 2022, they become strongly linked with perceived gaps, degree interest, and job anxiety.
Pre‑AI Period (Before November 2022)
- In the pre‑AI window, the gap conversation is already well‑defined. Knowledge gap and skill gap are strongly correlated (around 0.8), and leadership gap tracks closely with them (correlations around 0.6–0.7).
- Getting a degree sits inside this same complex, showing correlations in roughly the 0.6–0.8 range with the gap and leadership terms.
- Gen AI exists as a weak background signal with modest correlations; ChatGPT is effectively absent.
- Can’t find a job is positively related to the gap and degree series, but the system is still somewhat fragmented. The “job market” search term, in particular, often runs weakly or even negatively against some of the gap and anxiety terms. In other words, broad “job market” searches do not consistently move in sync with can’t find a job or the gap worries before AI takes off.
Before the AI breakout, there was an ongoing discussion about skills, knowledge, education, and job anxiety, but it was not strongly entangled with AI.
Post‑AI Period (November 2022 and Later)
Using the post‑AI correlation matrix:
- Gen AI and ChatGPT become central players. They show strong positive correlations with:
- knowledge gap
- skill gap
- leadership gap
- getting a degree
- In the post‑AI correlation matrix, these AI terms often have correlations above 0.6–0.7 with the capability and education series, indicating that when people pay attention to AI, they are simultaneously more likely to search about gaps and schooling.
- Can’t find a job is also strongly correlated with Gen AI and ChatGPT in the post‑AI period, and its correlations with skill gap, leadership gap, and getting a degree remain high. That is, personal job difficulty, AI interest, and perceived capability gaps increasingly move together.
- Job market flips from being loosely or even negatively related to some anxiety terms to being strongly positively correlated with:
- skill gap
- can’t find a job
- getting a degree
Overall, the system tightens:
- The correlations among AI, gap, education, and job‑market searches become uniformly strong and positive.
- Spikes in AI interest coincide with spikes in searches about whether people have the right skills, whether they should get more education, and how the job market is doing.
What This Means for Students, Workers, and Educators
Putting the findings together, the data points to a growing AI-shaped decision environment around work and education:
- There is a large, shared attention cycle where AI, skill gaps, education, and labor‑market anxiety largely move together, especially after late 2022.
- Before AI hit the mainstream, discussions about skill gaps, degrees, and job difficulty were present but less tightly connected to broad job‑market sentiment and AI topics.
- After Gen AI and ChatGPT broke through, nearly all of these signals become more synchronized:
- When AI interest rises, so do concerns about skill and knowledge gaps.
- People search more about degrees and about the job market in general.
The data suggests that AI is now embedded in how people think about employability and education choices. Concerns about skills, credentials, and job prospects are increasingly being interpreted through an AI-and-career-impact lens. It’s more important than ever for higher education to connect degree programs and workforce training to emerging technologies, changing skill expectations, and the real anxieties people have about how AI may affect their future opportunities.
Data and Methods
This study uses monthly Google Trends indices (0–100 scale, relative interest within each term) for the United States, from 2004 through early 2025, for seven topics:
- knowledge gap
- skill gap
- leadership gap
- Gen AI
- ChatGPT
- can’t find a job
- job market
- getting a degree
Key steps in the analysis:
- Converted all series to numeric form and constructed a clean monthly panel from 2004–2025.
- Computed Pearson correlations across all seven series to measure how often the topics move together.
- The core pairwise relationships emphasized in this report are statistically significant where measurable, with none of the headline correlations reported here exceeding p = .001.
- Built normalized time-series plots, scaling each series from 0–1 to focus on shared shapes rather than raw levels.
- Split the data into:
- Pre‑AI period: before 1 November 2022
- Post‑AI period: 1 November 2022 and later
- Because the post-AI period is substantially shorter than the pre-AI period, and because Google Trends reports relative rather than absolute search intensity, post-2022 findings should be interpreted as early descriptive evidence rather than causal proof or final long-run estimates.
- To address potential non-stationarity in the level series, we also repeated the analysis using month-over-month changes as a robustness check. These change-based correlations are, as expected, much smaller than the level-based estimates—generally around 0.1 to 0.2 rather than 0.5 to 0.8—but some remain statistically significant, suggesting the main findings are not driven only by shared long-run trends.
