Where Psychological Assessment Is Heading: What Assessment Clinicians Should Prepare For

Where Psychological Assessment Is Heading

Most writing about the future of assessment jumps to virtual reality and brain scans. The changes reaching a working assessment practice in the next year or two are less dramatic and more useful to plan for: how tests are administered and scored, how reports get drafted, and what you now need to tell clients and payers about the tools you use.

This guide covers six of those changes. For each one: what is happening, and what to do about it in your own workflow.

1. New Test Editions Are Built Around Digital Administration

Publishers are building newer editions around tablet administration and online scoring platforms. The WJ V is fully digital, with no physical kit, and is scored in Riverside Score. The WAIS-V can be given digitally as well as on paper. If you test regularly, your next edition is likely to come with a subscription to a scoring portal.

What this changes for you:

• Your score output format matters. Online scoring platforms produce score reports as PDFs, and some also offer data exports. Check what the output contains and how easily you can transfer scores into your report. Retyping a full battery's scores creates opportunities for transcription errors.
• Record retention moves online. If protocols live in a publisher's portal, check how long the publisher keeps them, how you export them, and what happens if you cancel your subscription. Your state's retention period for test records does not change because the protocol is digital.
• Plan the edition transition. Running two editions side by side for months makes templates, report language and peer review harder to manage. Plan how and when your practice will switch.

PsychReport keeps an instrument page for current editions, including the WJ-V and WAIS-V. You can upload the publisher's PDF score report to pre-fill scores for your review (how Smart Score Import works).

2. Remote and Hybrid Assessment Is Staying

Remote testing spread out of necessity and stayed because it serves rural clients, clients with transportation limits, and families who cannot take a day off for a clinic visit. It also brought documentation duties that in-person testing did not.

When planning a remote or hybrid evaluation:

• Follow instrument-level guidance. Remote administration guidance differs by instrument and subtest. Some measures are not appropriate to give remotely at all. Check the publisher's guidance before scheduling.
• Write the conditions into the report. State how testing was done (remote, in person or hybrid), who was present on the client's side, what equipment was used, and any departure from standardized administration. Explain how those conditions affect your confidence in the results.
• Check licensure across state lines. If the client is in a different state at the time of testing, your license and that state's rules govern the service. If you are in a PSYPACT state, review what the compact covers for telepsychology before relying on it.

3. AI Is Moving Into the Report-Writing Step

The most immediate change is in documentation. A complex evaluation can take 8 to 10 hours to write up by hand, and a typical report takes 4 to 6 hours. Tools built for assessment reports now draft the narrative from the scores, background information and observations you enter, so the clinician starts from a draft instead of a blank page.

Clinical responsibility stays with you: choosing the battery, administering and scoring it with the publisher's tools, judging validity, integrating the history, reaching the diagnosis and signing the report. A drafting tool should handle score entry and narrative drafting. It should not score tests, make a diagnosis or decide what the findings mean.

Review AI drafts for these recurring errors:

• Scores. Check every standard score, percentile and classification label against the score report. One transposed number in a summary table can change an eligibility decision.
• Identity details. Check names, pronouns, ages, grades and testing dates, especially if you started from an older template.
• Unsupported statements. Look for any history, observation or conclusion that you did not provide. If it is not in your notes, it does not belong in the report.
• Hedging and certainty. Make sure the strength of each conclusion matches your evidence. Drafts can sound more certain, or more vague, than you would be.
• Recommendations. Confirm that each recommendation follows from the findings and fits the client's setting (school, home, workplace).

If a tool makes this review hard, for example by not showing which scores it used, treat that as a reason to look elsewhere. Our guide to using AI for psychological assessment explains the review step in more detail.

4. Billing and Documentation Rules Apply to AI-Assisted Reports

Report preparation is part of the psychological evaluation services codes (96130 for the first hour and 96131 for each additional hour), which also cover integrating data, interpretation and feedback. These codes are time-based. If AI reduces your writing time, bill the time you actually spend reviewing, integrating and editing, not the time the report would have taken by hand.

Keep a time log for each evaluation that separates test administration, integration and writing, and feedback. Also keep notes on what you changed in the draft, so you can show how the signed report reflects your judgment if a payer or board asks.

5. Ethics Guidance and State Laws Are Catching Up

For several years the profession had little formal guidance on AI. In 2025 that started to change:

• APA guidance. In June 2025, APA published Ethical Guidance for AI in the Professional Practice of Health Service Psychology. It covers transparency with clients, informed consent, data privacy, bias, accuracy, and the psychologist's continuing responsibility for work products that AI helped produce. We summarize the practical points in our APA ethics and AI guide.
• State law. Illinois enacted the Wellness and Oversight for Psychological Resources Act in August 2025 (announcement from the Illinois Department of Financial and Professional Regulation). It bars AI from providing therapy or making therapeutic decisions and sets conditions on how licensed professionals may use AI for administrative and supplementary support. Check whether each state where you practice has adopted similar rules.
• HIPAA has not changed, but the tools have. Consumer tiers of general-purpose AI tools do not include a Business Associate Agreement (BAA). Some vendors offer BAAs on business or API plans, so check before entering any client data. Our HIPAA buyer's guide for AI tools lists the questions to ask.

Add a plain-language explanation of AI-assisted documentation to your consent paperwork. For example: "I may use secure software that drafts parts of your report from the information I collect. I review and edit every part of the report, and I am responsible for its content." Adjust it to your state's rules and your board's expectations.

6. Adaptive and Short-Form Measures Are Spreading

Computerized adaptive testing selects each item based on earlier answers, allowing it to reach a reliable score with fewer items. It is already standard in some health outcome measures, such as the NIH-funded PROMIS item banks. More rating scales and screeners are also offered in adaptive or short forms.

In the report, identify whether you used an adaptive administration, a short form or the full-length scale. Report score precision as well. Adaptive scores usually come with a standard error, which indicates uncertainty around the score. A brief explanation helps readers avoid overinterpreting small differences.

What Is Still Mostly Research

Some technologies attract attention but are not ready to support conclusions in a clinical report:

• Virtual reality tasks. VR continuous performance tests and virtual-environment memory tasks have research behind them. Check the norm sample, validity evidence and payer acceptance before citing one in a report.
• Passive data from phones and wearables. Sleep, activity and phone-use data may add context someday. For now, treat these data like other collateral information: describe their source and limits, and do not score them.
• Speech, video and facial analysis. Research tools can detect patterns in speech or video, but they are not validated diagnostic instruments. If a product claims otherwise, ask for peer-reviewed validation and details of the population it was tested on.

Before relying on a tool in a report, consider whether you could defend its use in a due process hearing or custody case. If you could not, it may inform your thinking, but it should not appear in the report as evidence.

A Short Checklist for This Year

1. Review the score output of your testing platforms and decide how scores will reach your reports.
2. Add a remote-administration paragraph to your report template that covers testing conditions and departures from standardization.
3. Create and use a review checklist for every AI-drafted report: scores, identity details, unsupported statements, certainty and recommendations.
4. Add an AI disclosure sentence to your consent forms and check your state's rules.
5. Confirm that a signed BAA is in place for every tool that touches client data.

Where PsychReport Fits

PsychReport turns scores and notes into a draft report. You enter scores or upload the publisher's score report, add background information and observations, and review, edit and sign the resulting narrative. It does not score tests or make diagnoses.

PsychReport supports 290+ assessments, with new ones added within 24 to 48 hours of a practitioner request. Practitioners save 4 or more hours per report.

The BAA is self-service and signed during onboarding. AI processing uses Zero Data Retention (ZDR), data is encrypted with TLS 1.3 in transit and AES-256 at rest, and the platform is hosted on SOC 2 Type II certified infrastructure. See security details or start with 3 free reports.

For an explanation of what goes into a report, see What Is a Psychological Report?.