For an Indian professional choosing what to learn next, a headline about AI skills can be useful and still describe the wrong job market. A count of technology advertisements, a collection spanning multiple roles and a survey of workers measure different things. Before turning any of them into a course plan, check who and what was counted.
The same label hides different samples
Qarera’s skills report covers 360,336 postings collected from December 27, 2025 to June 16, 2026. General AI appears in 19.8%; the specific label for prompt engineering appears in 1.6%. It extracts and standardises skills from descriptions and groups job families using title keywords. These are shares of postings, not completed hires. The report does not give a country breakdown, so its headline percentage cannot be presented as India’s hiring rate.
Its separate methodology page, updated October 17, describes an index built from employers’ applicant tracking systems and public job feeds. It warns that structured hiring systems are overrepresented, smaller or informal employers are underrepresented, and location classification and duplicate removal are imperfect. That explanation of the current index supplies useful cautions; it does not fill in the missing geography of the earlier skills sample.
Dice offers a more geographically explicit comparison. Its “Tech Hiring Myths vs. Reality” report says it analysed more than seven million US technology postings using Lightcast data pulled in May 2026. The share requesting at least one AI skill rose from 15% in January 2024 to 73% in May 2026. This is a measure of US technology postings, not a statistic for all jobs, all countries or Indian graduates. The report’s landing page does not specify the complete extraction window or reproduce its rules for classifying skills.
Build the shortlist around the work
The difference between a broad AI label and a named technique matters when choosing training. A job description could ask a product manager to evaluate an AI feature, a developer to integrate a model or a data specialist to maintain the information feeding it. Those are examples of different tasks, not evidence that one course prepares a candidate for all three.
A practical next step is to make a dated shortlist of genuine openings in the intended city, role and experience band. Separate required skills from preferred ones, check duplicate advertisements and record the actual task attached to each tool. Then compare that list with what a course teaches and what the candidate can demonstrate. This is a personal decision aid, not a substitute for a national survey of the labour market.
The Crunch
Research on job postings describes advertisements. It does not establish completed hires, prove that AI caused hiring, or identify a qualification that guarantees an offer. For an Indian applicant, the relevant job and its tasks matter more than an imported headline percentage.
Leave your thoughts