Assessment trends 2027: 10 shifts in how organizations test skills

The assessment trends that matter most for 2027 are a move from recall to applied skills, more continuous evidence, human oversight of AI marking, and firm regulatory dates for AI and accessibility. This guide covers ten shifts for training managers, L&D teams and certification owners. Each one has a dated source and a first move.

Every trend carries an evidence label: a rule with a date, a published standard, survey data, a regulator’s statement, or an expert or vendor view. Predictions no source measured are marked editorial. Facts are as of November 2026 and come from pages published between 2022 and July 2026.

Key takeaways

• Four of the ten shifts rest on a dated rule or published standard. The others rest on survey data, a regulator’s statement or expert views, and the labels below say which is which.
• What you test is changing first: from recalling facts toward applying skills, checked more than once.
• England’s exam regulator requires human judgment in high-stakes marking, and EU high-risk AI rules for some assessment uses are now scheduled for 2 December 2027.
• US public entities face WCAG 2.1 AA dates in April 2027 and April 2028.

Which assessment trends have the strongest evidence behind them?

Four trends rest on rules or standards with dates: the EU AI Act, the US ADA Title II web rule, QTI 3.0 and W3C Verifiable Credentials 2.0. One rests on survey data and one on a regulator’s statement. The other four rest on expert or vendor opinion, so treat them as direction.

Which assessment trends change what organizations test?

Four trends change the content of tests: competence is defined around applying and overseeing AI, evidence is gathered more than once, tasks replace some recall items, and hiring teams want verifiable skills. If you write or buy assessments, act on these first, because they change what an item should measure.

1. What counts as competence once AI can recall facts?

Competence is moving from remembering information toward applying it, judging it and overseeing automated systems. That is the view of Janet Garcia, CEO of PSI and President of ETS, in a January 2026 ETS article. She writes that “the more fundamental change is happening in the workplace,” and that AI literacy has no universal standard.

The article offers no data, so this is an expert view. The practical question is whether your exam rewards remembering a procedure or deciding what to do when the procedure does not fit. AI-assisted item writing is covered in our guide to how AI is changing test design.

2. Will continuous evidence replace the single exam?

Not soon, but the direction is clear. The same ETS article describes a move away from single-point multiple-choice exams toward “continuous, real-time demonstrations of competence,” and says certification will be less about proving competence once every few years. Garcia also expects slow change, citing the redesigned TOEFL’s long lead times.

Education Advanced, a vendor serving school districts, wrote in January 2026 that through-year testing is becoming the norm, without data. Our January 2024 post on what modern learners look for described the same pull toward ongoing upskilling from the learner’s side. Editorial prediction: expect short checks between formal exams well before exams disappear.

3. Are task-based items replacing multiple-choice questions?

Two assessment vendors say hands-on tasks are gaining ground, but neither measured how many programs have switched. TrueAbility wrote in July 2026 that “selecting the correct answer isn’t the same as performing the task,” and said generative AI pushes providers toward scenario-based testing. Questionmark made a similar case in December 2025 for observational testing in high-stakes and compliance settings.

Both sell assessment products and cite no independent study here. Editorial judgment: task items take more effort to build and score than multiple-choice items, so add them where the decision they support carries weight.

4. Do employers verify skills, or just say they do?

Both, and the sources agree once you separate adoption from execution. NACE’s Job Outlook 2026, published January 12, 2026, found that 70% of respondents use skill-based hiring, up from 65% the year before. Only 42% screen by GPA, down from 73% in 2019. NACE does not report a figure for skills assessments specifically.

The sample is small: 183 employers, surveyed between August 7 and September 22, 2025. Fortune reported on March 2, 2026 that 53% of employers say their organizations lack standardized hiring practices, citing a University of Phoenix survey of 2,000 US hiring stakeholders and job seekers (year not stated). Trust NACE for direction and Phoenix for the caution. If you run skills tests for hiring, the opening is standardized, scored tasks rather than self-ratings.

Pro tip: Tag Items By Recall Or Application. Before redesigning anything, tag every item on your highest-stakes exam as recall or application. The tagged list shows where a scenario item would change what the exam measures and where it would only add cost.

Which assessment trends change how AI is governed in testing?

Two trends set the rules for AI in testing: England’s regulator insists on human judgment in high-stakes marking, and the EU AI Act lists some assessment uses as high-risk, with an application date now set at 2 December 2027. Both change how you document and review automated scoring.

5. Can AI mark high-stakes tests on its own?

Not under Ofqual’s current rules. In a January 14, 2026 blog post, Ofqual said using AI as the only way to award marks does not meet its requirement for human judgment in marking decisions. It called AI promising for quality assurance and marker training but “nowhere near ready” to take over high-stakes marking.

Ofqual named “bias, inaccuracies and a lack of transparency” as fairness risks. Our January 2024 post on AI in assessments listed bias and limited contextual understanding as open challenges; now regulators name them too. The e-Assessment Association’s January 2026 report presents automation combined with human moderation as the best approach. Ofqual regulates qualifications in England, so check which regulator applies to you.

6. Does the EU AI Act cover assessment tools?

Some of them. Annex III of the AI Act lists AI systems intended “to evaluate learning outcomes” and for “monitoring and detecting prohibited behaviour of students during tests” among high-risk uses. It also lists systems intended for “the recruitment or selection of natural persons.” Whether a given tool is in scope depends on its intended purpose.

The Digital Omnibus on AI, Regulation (EU) 2026/1744, entered into force on 27 July 2026. A National Law Review summary reports that it moves the Annex III date from 2 August 2026 to 2 December 2027. This is general information, not legal advice; check with counsel.

Which assessment trends change how tests are delivered and shared?

Four trends affect delivery: trust-based integrity, accessibility deadlines, open standards for moving test content, and verifiable credentials for sharing results. They matter most when you sell exams to clients or move results between systems, because each depends on what your platform can export and document.

7. How will exam integrity work in 2027?

Integrity is shifting from catching cheaters toward building trust. The eAA report calls trust “the defining issue of this next phase of transformation” and lists privacy by design, risk-based monitoring and human review as themes. Garcia expects AI to enable large item pools and near-unique exam forms, and expects behavioral signals to help detect impersonation. Those are expectations, not results. Editorial read: with AI test monitoring named as high-risk in trend 6, expect monitoring that is proportionate, documented and reviewed by a person.

8. When do accessibility rules start to apply?

For US state and local governments, the WCAG 2.1 Level AA dates are April 26, 2027 for entities with populations of 50,000 or more, and April 26, 2028 for smaller entities and special district governments. The Department of Justice moved them in an interim final rule published April 20, 2026.

The original dates were April 24, 2026 and April 26, 2027. The rule invited comments, so check for updates. Whether an assessment portal run for a public client is covered is a question for counsel.

9. Why do open standards matter more now?

Standards decide whether your items and results can move between tools. 1EdTech lists QTI 3.0 as a final release dated 11 May 2022, merging earlier QTI and APIP into one specification with more native support for computer adaptive testing. Its results-reporting specifications are listed as candidates, not final releases.

Education Advanced also calls data fragmentation a growing bottleneck, without data. Editorial prediction: buyers will ask for exportable items and results more often. No source we read measured it.

10. What are verifiable credentials, and why should assessors care?

A verifiable credential is “a set of claims made by an issuer about a subject,” in W3C’s wording. The holder can present it, and a verifier can check that it came from the issuer and was not altered. W3C published Verifiable Credentials 2.0 as a standard on 15 May 2025.

In that announcement, RANDA Solutions said the specifications informed 1EdTech’s CLR 2.0 and Open Badges 3.0. Editorial read: a pass result could become a credential others verify without contacting you. The sources we read had no adoption data.

Pro tip: Ask For Sample Exports. Before choosing or renewing an assessment platform, ask the vendor for a sample item export, a sample results export, and an explanation of how a credential can be verified outside the platform.

Which assessment trends should you act on first?

Start with trends that have dates, then those that change what you test. These priorities are editorial judgments built from the evidence labels above, not study findings. A deadline outranks an expert prediction, and a change to what you test comes before a change to how you deliver it.

What should you do next?

Pick one exam and run the recall-or-application tag from the first pro tip. Then send your platform vendor the export questions from the second. Together they take an afternoon.

SimpliAssess’s published pages describe a headless, API-first platform with item banking, analytics and reporting, and a white-label option. If standards such as QTI or verifiable credentials matter to you, ask what is supported today in a SimpliAssess demo.

For the look back at what actually changed this year, read our review of online assessment in 2026.

Frequently asked questions

What are the main assessment trends for 2027?

The ten covered here are applied competence, continuous evidence, task-based items, verified skills in hiring, human oversight of AI marking, EU AI Act rules, trust-based integrity, accessibility deadlines, open standards and verifiable credentials. Four rest on dated rules or standards; the others vary from survey data to expert opinion.

Will AI replace human markers?

For high-stakes exams in England, not under current rules. Ofqual said in January 2026 that using AI as the only way to award marks does not meet its requirement for human judgment. It sees promise in quality assurance and marker training. Other regulators may differ, so check the rules that apply to you.

Are multiple-choice tests going away?

No source we read says they will disappear. ETS describes a movement away from single-point multiple-choice exams toward more continuous demonstrations of competence. Editorial view: keep multiple-choice items for recall checks, and add task or scenario items where decisions carry weight.

Does the EU AI Act apply to assessment software?

It can. Annex III lists AI systems intended to evaluate learning outcomes, to monitor student behavior during tests, and to recruit or select people as high-risk uses. A National Law Review summary of Regulation (EU) 2026/1744 gives the new application date as 2 December 2027. Scope depends on a tool’s intended purpose, so ask counsel.

What is the difference between skills-based hiring and skills testing?

Skills-based hiring is the broader practice of judging candidates on skills rather than degrees. A skills test is one way to measure those skills, alongside structured interviews and work samples. NACE’s 2026 survey reports 70% of respondents use skill-based hiring but gives no figure for skills assessments specifically.