UCLA AI Scribe Study 2026: What Physicians Need to Know

Updated January 2026

A landmark UCLA Health study published in the New England Journal of Medicine AI examined the real-world effectiveness of AI scribe technology across 238 physicians, 14 specialties, and 72,000 patient encounters. The results provide critical evidence for physicians considering AI documentation tools.

Key Finding: Nabla AI scribe users reduced their documentation time by approximately 41 seconds per note (nearly 10%), with physicians reporting potential benefits for burnout and work-related stress.

This guide breaks down the study findings, explains what they mean for practicing physicians, and provides practical guidance on implementing AI scribe technology based on evidence.

Create Your AI-Powered Documentation in 2 Minutes

Start with 20 free SOAP notes. No credit card required.

Try Free on Web

Study Overview: What UCLA Examined

Research Design

Scope:

238 physicians across 14 specialties
72,000 patient encounters
6-month study period
Two AI scribe tools: Microsoft DAX and Nabla
Control group using usual documentation methods

Specialties Included:

Primary Care (Family Medicine, Internal Medicine)
Surgical Specialties
Medical Specialties (Cardiology, Pulmonology, etc.)
Pediatrics
Emergency Medicine
Psychiatry
And 8 other specialties

Outcomes Measured:

Documentation time per note
Physician burnout scores
Work-related stress levels
Documentation accuracy and completeness
Patient safety events

Key Findings at a Glance

✅ Time Savings: Nabla users saved ~41 seconds per note (10% reduction) ✅ Burnout Reduction: Potential benefits for burnout and work-related stress ✅ Multi-Specialty Effectiveness: Benefits observed across 14 different specialties ⚠️ Accuracy Concerns: Clinically significant inaccuracies noted "occasionally" ⚠️ Safety Event: One mild patient safety event reported ✅ Provider Satisfaction: Physicians reported overall positive experience

Detailed Results: What the Numbers Show

Documentation Time Reduction

Nabla AI Scribe Results:

Before: 4 minutes 30 seconds average per note
After: 3 minutes 49 seconds average per note
Time Saved: 41 seconds per note (9.1% reduction)

Cumulative Daily Impact: For a physician seeing 20 patients per day:

Time saved: 41 seconds × 20 patients = 13.7 minutes daily
Weekly savings: ~1 hour 8 minutes per week
Annual savings: ~60 hours per year

Microsoft DAX Results:

Also showed documentation time benefits
Specific time savings varied by specialty and user
Overall positive trend but less dramatic than Nabla

Burnout and Well-Being Impact

The study found "potential benefits for physician burnout and work-related stress" among AI scribe users. This aligns with other research:

Complementary Evidence:

JAMA Network Open study: 31% drop in reported burnout with ambient AI scribes
Oncology Nurse Advisor report: AI scribes may reduce documentation time and burnout

Mechanism of Burnout Reduction:

Less after-hours documentation work
More time with patients during encounters
Reduced cognitive load from documentation tasks
Increased sense of control over workday

Accuracy and Safety Findings

Critical Finding: The study noted clinically significant inaccuracies appeared "occasionally" on 5-point Likert scale assessments.

Types of Inaccuracies Identified:

Omissions: Missing clinical information
Pronoun errors: Incorrect attribution of symptoms or history
Terminology mistakes: Incorrect medical terms or anatomical locations
Context errors: Misunderstanding of clinical context

Patient Safety Event:

One mild patient safety event reported during the study
Event was identified and corrected through physician review
Underscores importance of verification before signing notes

Safety Implications: ⚠️ Never sign AI-generated notes without thorough review ⚠️ Verify all diagnoses, medications, and treatment plans ⚠️ Check that clinical context is accurately captured ⚠️ Ensure all patient-reported symptoms are correctly documented

What This Means for Your Practice

Should You Adopt AI Scribe Technology?

The Evidence Supports AI Scribes If:

✅ You spend >2 hours daily on documentation ✅ You regularly document after clinic hours ✅ You feel documentation detracts from patient care ✅ You're experiencing burnout or work-related stress ✅ You're willing to implement structured review processes

Consider Waiting If:

❌ You're already efficient with documentation (<30 min/day) ❌ Your organization doesn't support AI tool adoption ❌ You're uncomfortable with technology implementation ❌ Budget constraints prevent investment in AI tools ❌ Your specialty has unique documentation that AI may not handle well

Which AI Scribe Tool Should You Choose?

The UCLA study examined Microsoft DAX and Nabla, but many other options exist in 2026:

Tools with Published Efficacy Data:

Nabla: UCLA study showed 10% documentation time reduction
Microsoft DAX (Nuance): Included in UCLA study, widely adopted
Abridge: Growing body of evidence for effectiveness

Other Notable Options:

OpenAI for Healthcare: Launched January 2026 (learn more)
athenaAmbient: Launching February 2026, free for athenahealth users
SOAPNoteAI: Purpose-built SOAP note generation with ambient listening

Selection Criteria:

Evidence: Does it have peer-reviewed efficacy data?
Specialty fit: Does it support your specialty's documentation needs?
EHR integration: Does it work with your electronic health record?
HIPAA compliance: Does it offer Business Associate Agreements (BAAs)?
Cost: Does the pricing fit your budget?
Support: What implementation and ongoing support is provided?

Implementing AI Scribes: Lessons from UCLA Study

Best Practices for Implementation

Based on the UCLA study findings and clinical experience:

1. Start with a Pilot Program

Select 2-5 physicians to test AI scribe first
Choose early adopters comfortable with technology
Run pilot for 4-8 weeks before broader rollout
Gather structured feedback weekly

2. Establish Review Protocols Given the "occasional" inaccuracies found in the UCLA study:

Mandatory Review Checklist:

Verify all diagnoses are accurate
Check all medications (names, doses, frequencies)
Confirm procedures and treatments are correctly documented
Ensure patient-reported symptoms match your clinical assessment
Verify all anatomical terms and laterality (left/right)
Check for pronoun errors (patient vs. family member statements)
Confirm all follow-up plans and referrals are accurate

3. Train Physicians on Effective AI Use

Conversation Techniques for Better AI Capture:

Speak clearly and use complete sentences
Explicitly state clinical findings ("The patient has...")
Verbalize key negative findings
State diagnoses and treatment plans explicitly
Avoid ambiguous pronouns ("he said" - specify who)

Example: Good vs. Poor AI Scribe Input

Poor (Ambiguous):

"So the pain is worse at night?" "Yeah." "Okay, and it radiates?" "Sometimes."

Good (Clear for AI):

"The patient reports that his lower back pain worsens at night. The pain occasionally radiates down his left leg to the knee, consistent with L5 radiculopathy."

4. Monitor Quality Metrics

Track These Metrics:

Documentation time per note (before vs. after)
Edit rate (% of AI-generated text modified)
Documentation completeness scores
Billing coding accuracy
Physician satisfaction surveys
Patient safety events related to documentation

Target Benchmarks:

Documentation time reduction: 15-30%
Edit rate: <25% (lower is better, but some editing is expected)
Documentation completeness: >90%
Physician satisfaction: >80% positive
Safety events: Zero tolerance

Common Pitfalls to Avoid

Based on UCLA study and clinical experience:

1. Over-Reliance on AI ❌ Don't assume AI is always correct ✅ Always verify before signing

2. Insufficient Training ❌ Don't deploy without proper physician training ✅ Provide hands-on training and support

3. Poor Audio Quality ❌ Don't use AI scribe in noisy environments without optimization ✅ Use quality microphones and test audio capture

4. Ignoring Edit Patterns ❌ Don't ignore repeated AI errors ✅ Report systematic errors to vendor for improvement

5. Rushing Implementation ❌ Don't force adoption without buy-in ✅ Let early adopters demonstrate value first

Comparing UCLA Study to Other Research

Supporting Evidence for AI Scribes

JAMA Network Open Multicenter Study (2024-2025):

Finding: 31% drop in reported burnout
Finding: 30% boost in physician well-being
Scope: Multicenter randomized controlled trial
Conclusion: Ambient AI scribes significantly improve well-being

VA National Pilot (2025-2026):

Finding: VA piloted ambient AI with 800,000+ veterans
Finding: Nationwide rollout planned for 2026
Significance: Largest government healthcare AI deployment
Source: Military.com report

GeneOnline Review (2025):

Finding: AI scribes consistently save 1-2 hours documentation time daily
Finding: ~20% cuts in note-taking time typical
Finding: ~30% reductions in after-hours work
Conclusion: Technology is mature and effective

Addressing Skepticism

Common Physician Concerns:

Concern 1: "AI documentation isn't accurate enough." Evidence: UCLA study showed inaccuracies are "occasional" and caught through review. No serious safety events occurred in 72,000 encounters.

Concern 2: "It will take more time to edit AI notes than write my own." Evidence: UCLA study showed net 10% time savings even after editing. Other studies show 20-30% savings.

Concern 3: "I don't trust AI with my patients." Response: AI assists, but you remain in control. You review and approve everything before it becomes part of the medical record.

Concern 4: "My documentation style is too specialized." Evidence: UCLA study included 14 different specialties with positive results across all of them.

Financial Analysis: Is AI Scribe Worth the Cost?

Cost-Benefit Calculation

Typical AI Scribe Costs (2026):

Individual subscription: $100-200/month
Enterprise deployment: $150-300/physician/month
Implementation costs: $500-2,000 one-time

Time Value Calculation: Using UCLA study findings (41 seconds per note):

Scenario: Physician seeing 20 patients/day

Time saved: 13.7 minutes/day
Annual time saved: ~60 hours
Value of time (at $150/hour): $9,000/year
Annual tool cost: $1,200-2,400/year
Net benefit: $6,600-7,800/year

Additional Financial Benefits:

Billing optimization: 10-15% revenue increase from better documentation
Reduced burnout: Lower turnover costs
Increased patient volume: See more patients with saved time

ROI Timeline: Typically 3-6 months to positive return

Billing and Coding Impact

The UCLA study didn't specifically examine billing impacts, but clinical experience shows:

Documentation Completeness → Higher Billing Levels:

Better documentation supports higher E/M codes
AI prompts for missing elements needed for billing levels
Average billing increase: 10-20% where clinically appropriate

Example:

20 patients/day × 250 work days/year = 5,000 encounters/year
10% billing optimization × $150 average/encounter = $75,000 additional revenue
AI scribe cost: $2,400/year
Net billing benefit: $72,600/year

Specialty-Specific Considerations

Specialties Where AI Scribes Excel

Based on UCLA study and clinical feedback:

High-Value Specialties:

Primary Care: High visit volume, straightforward documentation
Psychiatry/Mental Health: Conversational encounters translate well
Pediatrics: Parent conversations captured effectively
Urgent Care: High volume, time-pressure benefits
Cardiology: Complex but structured assessments

Specialties with Unique Challenges

May Require Specialty-Specific Training:

Surgery: Operative notes have specific structure and terminology
Emergency Medicine: Rapid encounters, interruptions common
Radiology: Findings-based documentation, less conversational
Pathology: Microscopic findings, technical terminology

These specialties can still benefit but may need:

Specialty-specific templates
Additional training on verbalization of findings
Custom workflows for procedure documentation

The Future: Beyond the UCLA Study

Next Generation AI Scribe Features

Based on 2026 trends:

Emerging Capabilities:

Agentic AI: Beyond transcription to workflow management (learn more)
Multi-modal capture: Video, images, and audio integrated
Real-time guidance: AI suggests questions during encounter
Predictive documentation: AI anticipates likely documentation based on chief complaint

Coming Soon:

Autonomous order placement: AI generates orders for approval
Billing optimization: Real-time coding suggestions
Population health integration: AI links documentation to population metrics
Patient-facing summaries: Automatic after-visit summary generation

Regulatory Landscape

FDA Oversight:

AI scribes currently considered clinical decision support
Regulatory framework evolving in 2026
Some tools may require FDA clearance as AI capabilities expand

Reimbursement:

Medicare/Medicaid don't yet reimburse for AI scribe tools directly
Some commercial payers exploring coverage
Tax deductible as business expense

Conclusion: What Should You Do?

Action Steps Based on UCLA Study Evidence

If You're Considering AI Scribe Technology:

✅ Step 1: Review the UCLA study full text (NEJM AI) ✅ Step 2: Request demos from 2-3 AI scribe vendors ✅ Step 3: Start with a pilot program (4-8 weeks) ✅ Step 4: Establish mandatory review protocols ✅ Step 5: Measure outcomes (time, satisfaction, accuracy)

Key Takeaways from UCLA Study:

Time savings are real but modest: 10% reduction (41 seconds/note)
Burnout benefits are meaningful: Improved physician well-being
Accuracy requires vigilance: Always review AI-generated notes
Multi-specialty effectiveness: Works across diverse clinical settings
Safety events are rare: One mild event in 72,000 encounters

The Bottom Line:

The UCLA study provides high-quality evidence that AI scribe technology:

Reduces documentation time
Improves physician well-being
Works across multiple specialties
Is safe when used with appropriate oversight

For physicians struggling with documentation burden and burnout, AI scribes represent an evidence-based solution worth serious consideration.

Frequently Asked Questions

The UCLA study, published in the New England Journal of Medicine AI in 2026, examined 238 physicians across 14 specialties and 72,000 patient encounters. It found that Nabla users reduced documentation time by nearly 10% (41 seconds per note), with potential benefits for physician burnout and work-related stress.

The UCLA study examined two commercially available AI scribe applications: Microsoft DAX and Nabla. Both showed documentation benefits, with Nabla users seeing the most significant reduction in documentation time compared to usual care.

The study noted that clinically significant inaccuracies appeared 'occasionally' in AI-generated notes on 5-point Likert scale questions. One mild patient safety event was reported during the study. Physicians must always review and verify AI-generated documentation before signing.

Nabla users in the UCLA study saved approximately 41 seconds per note on average, reducing documentation time from 4 minutes 30 seconds to 3 minutes 49 seconds. Over a full clinical day with 20 patients, this translates to about 14 minutes saved daily.

Yes, the UCLA study found potential benefits for physician burnout and work-related stress among users of AI scribe technology. This aligns with a JAMA Network Open multicenter study that found a 31% drop in reported burnout with ambient AI scribes.

The UCLA study supports the use of AI scribes but emphasizes the critical importance of physician review. AI documentation should always be verified for accuracy, completeness, and clinical appropriateness before being finalized. The provider remains ultimately responsible for all documentation.

Medical Disclaimer: This content is for educational purposes only and should not replace professional medical judgment. Always consult current clinical guidelines and your institution's policies.

UCLA AI Scribe Study 2026: What Physicians Need to Know

Updated January 2026

This guide breaks down the study findings, explains what they mean for practicing physicians, and provides practical guidance on implementing AI scribe technology based on evidence.

Create Your AI-Powered Documentation in 2 Minutes

Start with 20 free SOAP notes. No credit card required.

Try Free on Web

Study Overview: What UCLA Examined

Research Design

Scope:

238 physicians across 14 specialties
72,000 patient encounters
6-month study period
Two AI scribe tools: Microsoft DAX and Nabla
Control group using usual documentation methods

Specialties Included:

Primary Care (Family Medicine, Internal Medicine)
Surgical Specialties
Medical Specialties (Cardiology, Pulmonology, etc.)
Pediatrics
Emergency Medicine
Psychiatry
And 8 other specialties

Outcomes Measured:

Documentation time per note
Physician burnout scores
Work-related stress levels
Documentation accuracy and completeness
Patient safety events

Key Findings at a Glance

Detailed Results: What the Numbers Show

Documentation Time Reduction

Nabla AI Scribe Results:

Before: 4 minutes 30 seconds average per note
After: 3 minutes 49 seconds average per note
Time Saved: 41 seconds per note (9.1% reduction)

Cumulative Daily Impact: For a physician seeing 20 patients per day:

Time saved: 41 seconds × 20 patients = 13.7 minutes daily
Weekly savings: ~1 hour 8 minutes per week
Annual savings: ~60 hours per year

Microsoft DAX Results:

Also showed documentation time benefits
Specific time savings varied by specialty and user
Overall positive trend but less dramatic than Nabla

Burnout and Well-Being Impact

The study found "potential benefits for physician burnout and work-related stress" among AI scribe users. This aligns with other research:

Complementary Evidence:

JAMA Network Open study: 31% drop in reported burnout with ambient AI scribes
Oncology Nurse Advisor report: AI scribes may reduce documentation time and burnout

Mechanism of Burnout Reduction:

Less after-hours documentation work
More time with patients during encounters
Reduced cognitive load from documentation tasks
Increased sense of control over workday

Accuracy and Safety Findings

Critical Finding: The study noted clinically significant inaccuracies appeared "occasionally" on 5-point Likert scale assessments.

Types of Inaccuracies Identified:

Omissions: Missing clinical information
Pronoun errors: Incorrect attribution of symptoms or history
Terminology mistakes: Incorrect medical terms or anatomical locations
Context errors: Misunderstanding of clinical context

Patient Safety Event:

One mild patient safety event reported during the study
Event was identified and corrected through physician review
Underscores importance of verification before signing notes

What This Means for Your Practice

Should You Adopt AI Scribe Technology?

The Evidence Supports AI Scribes If:

Consider Waiting If:

Which AI Scribe Tool Should You Choose?

The UCLA study examined Microsoft DAX and Nabla, but many other options exist in 2026:

Tools with Published Efficacy Data:

Nabla: UCLA study showed 10% documentation time reduction
Microsoft DAX (Nuance): Included in UCLA study, widely adopted
Abridge: Growing body of evidence for effectiveness

Other Notable Options:

OpenAI for Healthcare: Launched January 2026 (learn more)
athenaAmbient: Launching February 2026, free for athenahealth users
SOAPNoteAI: Purpose-built SOAP note generation with ambient listening

Selection Criteria:

Evidence: Does it have peer-reviewed efficacy data?
Specialty fit: Does it support your specialty's documentation needs?
EHR integration: Does it work with your electronic health record?
HIPAA compliance: Does it offer Business Associate Agreements (BAAs)?
Cost: Does the pricing fit your budget?
Support: What implementation and ongoing support is provided?

Implementing AI Scribes: Lessons from UCLA Study

Best Practices for Implementation

Based on the UCLA study findings and clinical experience:

1. Start with a Pilot Program

Select 2-5 physicians to test AI scribe first
Choose early adopters comfortable with technology
Run pilot for 4-8 weeks before broader rollout
Gather structured feedback weekly

2. Establish Review Protocols Given the "occasional" inaccuracies found in the UCLA study:

Mandatory Review Checklist:

Verify all diagnoses are accurate
Check all medications (names, doses, frequencies)
Confirm procedures and treatments are correctly documented
Ensure patient-reported symptoms match your clinical assessment
Verify all anatomical terms and laterality (left/right)
Check for pronoun errors (patient vs. family member statements)
Confirm all follow-up plans and referrals are accurate

3. Train Physicians on Effective AI Use

Conversation Techniques for Better AI Capture:

Speak clearly and use complete sentences
Explicitly state clinical findings ("The patient has...")
Verbalize key negative findings
State diagnoses and treatment plans explicitly
Avoid ambiguous pronouns ("he said" - specify who)

Example: Good vs. Poor AI Scribe Input

Poor (Ambiguous):

"So the pain is worse at night?" "Yeah." "Okay, and it radiates?" "Sometimes."

Good (Clear for AI):

"The patient reports that his lower back pain worsens at night. The pain occasionally radiates down his left leg to the knee, consistent with L5 radiculopathy."

4. Monitor Quality Metrics

Track These Metrics:

Documentation time per note (before vs. after)
Edit rate (% of AI-generated text modified)
Documentation completeness scores
Billing coding accuracy
Physician satisfaction surveys
Patient safety events related to documentation

Target Benchmarks:

Documentation time reduction: 15-30%
Edit rate: <25% (lower is better, but some editing is expected)
Documentation completeness: >90%
Physician satisfaction: >80% positive
Safety events: Zero tolerance

Common Pitfalls to Avoid

Based on UCLA study and clinical experience:

1. Over-Reliance on AI ❌ Don't assume AI is always correct ✅ Always verify before signing

2. Insufficient Training ❌ Don't deploy without proper physician training ✅ Provide hands-on training and support

3. Poor Audio Quality ❌ Don't use AI scribe in noisy environments without optimization ✅ Use quality microphones and test audio capture

4. Ignoring Edit Patterns ❌ Don't ignore repeated AI errors ✅ Report systematic errors to vendor for improvement

5. Rushing Implementation ❌ Don't force adoption without buy-in ✅ Let early adopters demonstrate value first

Comparing UCLA Study to Other Research

Supporting Evidence for AI Scribes

JAMA Network Open Multicenter Study (2024-2025):

Finding: 31% drop in reported burnout
Finding: 30% boost in physician well-being
Scope: Multicenter randomized controlled trial
Conclusion: Ambient AI scribes significantly improve well-being

VA National Pilot (2025-2026):

Finding: VA piloted ambient AI with 800,000+ veterans
Finding: Nationwide rollout planned for 2026
Significance: Largest government healthcare AI deployment
Source: Military.com report

GeneOnline Review (2025):

Finding: AI scribes consistently save 1-2 hours documentation time daily
Finding: ~20% cuts in note-taking time typical
Finding: ~30% reductions in after-hours work
Conclusion: Technology is mature and effective

Addressing Skepticism

Common Physician Concerns:

Concern 2: "It will take more time to edit AI notes than write my own." Evidence: UCLA study showed net 10% time savings even after editing. Other studies show 20-30% savings.

Concern 3: "I don't trust AI with my patients." Response: AI assists, but you remain in control. You review and approve everything before it becomes part of the medical record.

Concern 4: "My documentation style is too specialized." Evidence: UCLA study included 14 different specialties with positive results across all of them.

Financial Analysis: Is AI Scribe Worth the Cost?

Cost-Benefit Calculation

Typical AI Scribe Costs (2026):

Individual subscription: $100-200/month
Enterprise deployment: $150-300/physician/month
Implementation costs: $500-2,000 one-time

Time Value Calculation: Using UCLA study findings (41 seconds per note):

Scenario: Physician seeing 20 patients/day

Time saved: 13.7 minutes/day
Annual time saved: ~60 hours
Value of time (at $150/hour): $9,000/year
Annual tool cost: $1,200-2,400/year
Net benefit: $6,600-7,800/year

Additional Financial Benefits:

Billing optimization: 10-15% revenue increase from better documentation
Reduced burnout: Lower turnover costs
Increased patient volume: See more patients with saved time

ROI Timeline: Typically 3-6 months to positive return

Billing and Coding Impact

The UCLA study didn't specifically examine billing impacts, but clinical experience shows:

Documentation Completeness → Higher Billing Levels:

Better documentation supports higher E/M codes
AI prompts for missing elements needed for billing levels
Average billing increase: 10-20% where clinically appropriate

Example:

20 patients/day × 250 work days/year = 5,000 encounters/year
10% billing optimization × $150 average/encounter = $75,000 additional revenue
AI scribe cost: $2,400/year
Net billing benefit: $72,600/year

Specialty-Specific Considerations

Specialties Where AI Scribes Excel

Based on UCLA study and clinical feedback:

High-Value Specialties:

Primary Care: High visit volume, straightforward documentation
Psychiatry/Mental Health: Conversational encounters translate well
Pediatrics: Parent conversations captured effectively
Urgent Care: High volume, time-pressure benefits
Cardiology: Complex but structured assessments

Specialties with Unique Challenges

May Require Specialty-Specific Training:

Surgery: Operative notes have specific structure and terminology
Emergency Medicine: Rapid encounters, interruptions common
Radiology: Findings-based documentation, less conversational
Pathology: Microscopic findings, technical terminology

These specialties can still benefit but may need:

Specialty-specific templates
Additional training on verbalization of findings
Custom workflows for procedure documentation

The Future: Beyond the UCLA Study

Next Generation AI Scribe Features

Based on 2026 trends:

Emerging Capabilities:

Agentic AI: Beyond transcription to workflow management (learn more)
Multi-modal capture: Video, images, and audio integrated
Real-time guidance: AI suggests questions during encounter
Predictive documentation: AI anticipates likely documentation based on chief complaint

Coming Soon:

Autonomous order placement: AI generates orders for approval
Billing optimization: Real-time coding suggestions
Population health integration: AI links documentation to population metrics
Patient-facing summaries: Automatic after-visit summary generation

Regulatory Landscape

FDA Oversight:

AI scribes currently considered clinical decision support
Regulatory framework evolving in 2026
Some tools may require FDA clearance as AI capabilities expand

Reimbursement:

Medicare/Medicaid don't yet reimburse for AI scribe tools directly
Some commercial payers exploring coverage
Tax deductible as business expense

Conclusion: What Should You Do?

Action Steps Based on UCLA Study Evidence

If You're Considering AI Scribe Technology:

Key Takeaways from UCLA Study:

Time savings are real but modest: 10% reduction (41 seconds/note)
Burnout benefits are meaningful: Improved physician well-being
Accuracy requires vigilance: Always review AI-generated notes
Multi-specialty effectiveness: Works across diverse clinical settings
Safety events are rare: One mild event in 72,000 encounters

The Bottom Line:

The UCLA study provides high-quality evidence that AI scribe technology:

Reduces documentation time
Improves physician well-being
Works across multiple specialties
Is safe when used with appropriate oversight

For physicians struggling with documentation burden and burnout, AI scribes represent an evidence-based solution worth serious consideration.

Frequently Asked Questions

Medical Disclaimer: This content is for educational purposes only and should not replace professional medical judgment. Always consult current clinical guidelines and your institution's policies.

UCLA AI Scribe Study 2026: What Physicians Need to Know

Updated January 2026

This guide breaks down the study findings, explains what they mean for practicing physicians, and provides practical guidance on implementing AI scribe technology based on evidence.

Create Your AI-Powered Documentation in 2 Minutes

Start with 20 free SOAP notes. No credit card required.

Try Free on Web

Study Overview: What UCLA Examined

Research Design

Scope:

238 physicians across 14 specialties
72,000 patient encounters
6-month study period
Two AI scribe tools: Microsoft DAX and Nabla
Control group using usual documentation methods

Specialties Included:

Primary Care (Family Medicine, Internal Medicine)
Surgical Specialties
Medical Specialties (Cardiology, Pulmonology, etc.)
Pediatrics
Emergency Medicine
Psychiatry
And 8 other specialties

Outcomes Measured:

Documentation time per note
Physician burnout scores
Work-related stress levels
Documentation accuracy and completeness
Patient safety events

Key Findings at a Glance

Detailed Results: What the Numbers Show

Documentation Time Reduction

Nabla AI Scribe Results:

Before: 4 minutes 30 seconds average per note
After: 3 minutes 49 seconds average per note
Time Saved: 41 seconds per note (9.1% reduction)

Cumulative Daily Impact: For a physician seeing 20 patients per day:

Time saved: 41 seconds × 20 patients = 13.7 minutes daily
Weekly savings: ~1 hour 8 minutes per week
Annual savings: ~60 hours per year

Microsoft DAX Results:

Also showed documentation time benefits
Specific time savings varied by specialty and user
Overall positive trend but less dramatic than Nabla

Burnout and Well-Being Impact

The study found "potential benefits for physician burnout and work-related stress" among AI scribe users. This aligns with other research:

Complementary Evidence:

JAMA Network Open study: 31% drop in reported burnout with ambient AI scribes
Oncology Nurse Advisor report: AI scribes may reduce documentation time and burnout

Mechanism of Burnout Reduction:

Less after-hours documentation work
More time with patients during encounters
Reduced cognitive load from documentation tasks
Increased sense of control over workday

Accuracy and Safety Findings

Critical Finding: The study noted clinically significant inaccuracies appeared "occasionally" on 5-point Likert scale assessments.

Types of Inaccuracies Identified:

Omissions: Missing clinical information
Pronoun errors: Incorrect attribution of symptoms or history
Terminology mistakes: Incorrect medical terms or anatomical locations
Context errors: Misunderstanding of clinical context

Patient Safety Event:

One mild patient safety event reported during the study
Event was identified and corrected through physician review
Underscores importance of verification before signing notes

What This Means for Your Practice

Should You Adopt AI Scribe Technology?

The Evidence Supports AI Scribes If:

Consider Waiting If:

Which AI Scribe Tool Should You Choose?

The UCLA study examined Microsoft DAX and Nabla, but many other options exist in 2026:

Tools with Published Efficacy Data:

Nabla: UCLA study showed 10% documentation time reduction
Microsoft DAX (Nuance): Included in UCLA study, widely adopted
Abridge: Growing body of evidence for effectiveness

Other Notable Options:

OpenAI for Healthcare: Launched January 2026 (learn more)
athenaAmbient: Launching February 2026, free for athenahealth users
SOAPNoteAI: Purpose-built SOAP note generation with ambient listening

Selection Criteria:

Evidence: Does it have peer-reviewed efficacy data?
Specialty fit: Does it support your specialty's documentation needs?
EHR integration: Does it work with your electronic health record?
HIPAA compliance: Does it offer Business Associate Agreements (BAAs)?
Cost: Does the pricing fit your budget?
Support: What implementation and ongoing support is provided?

Implementing AI Scribes: Lessons from UCLA Study

Best Practices for Implementation

Based on the UCLA study findings and clinical experience:

1. Start with a Pilot Program

Select 2-5 physicians to test AI scribe first
Choose early adopters comfortable with technology
Run pilot for 4-8 weeks before broader rollout
Gather structured feedback weekly

2. Establish Review Protocols Given the "occasional" inaccuracies found in the UCLA study:

Mandatory Review Checklist:

Verify all diagnoses are accurate
Check all medications (names, doses, frequencies)
Confirm procedures and treatments are correctly documented
Ensure patient-reported symptoms match your clinical assessment
Verify all anatomical terms and laterality (left/right)
Check for pronoun errors (patient vs. family member statements)
Confirm all follow-up plans and referrals are accurate

3. Train Physicians on Effective AI Use

Conversation Techniques for Better AI Capture:

Speak clearly and use complete sentences
Explicitly state clinical findings ("The patient has...")
Verbalize key negative findings
State diagnoses and treatment plans explicitly
Avoid ambiguous pronouns ("he said" - specify who)

Example: Good vs. Poor AI Scribe Input

Poor (Ambiguous):

"So the pain is worse at night?" "Yeah." "Okay, and it radiates?" "Sometimes."

Good (Clear for AI):

"The patient reports that his lower back pain worsens at night. The pain occasionally radiates down his left leg to the knee, consistent with L5 radiculopathy."

4. Monitor Quality Metrics

Track These Metrics:

Documentation time per note (before vs. after)
Edit rate (% of AI-generated text modified)
Documentation completeness scores
Billing coding accuracy
Physician satisfaction surveys
Patient safety events related to documentation

Target Benchmarks:

Documentation time reduction: 15-30%
Edit rate: <25% (lower is better, but some editing is expected)
Documentation completeness: >90%
Physician satisfaction: >80% positive
Safety events: Zero tolerance

Common Pitfalls to Avoid

Based on UCLA study and clinical experience:

1. Over-Reliance on AI ❌ Don't assume AI is always correct ✅ Always verify before signing

2. Insufficient Training ❌ Don't deploy without proper physician training ✅ Provide hands-on training and support

3. Poor Audio Quality ❌ Don't use AI scribe in noisy environments without optimization ✅ Use quality microphones and test audio capture

4. Ignoring Edit Patterns ❌ Don't ignore repeated AI errors ✅ Report systematic errors to vendor for improvement

5. Rushing Implementation ❌ Don't force adoption without buy-in ✅ Let early adopters demonstrate value first

Comparing UCLA Study to Other Research

Supporting Evidence for AI Scribes

JAMA Network Open Multicenter Study (2024-2025):

Finding: 31% drop in reported burnout
Finding: 30% boost in physician well-being
Scope: Multicenter randomized controlled trial
Conclusion: Ambient AI scribes significantly improve well-being

VA National Pilot (2025-2026):

Finding: VA piloted ambient AI with 800,000+ veterans
Finding: Nationwide rollout planned for 2026
Significance: Largest government healthcare AI deployment
Source: Military.com report

GeneOnline Review (2025):

Finding: AI scribes consistently save 1-2 hours documentation time daily
Finding: ~20% cuts in note-taking time typical
Finding: ~30% reductions in after-hours work
Conclusion: Technology is mature and effective

Addressing Skepticism

Common Physician Concerns:

Concern 2: "It will take more time to edit AI notes than write my own." Evidence: UCLA study showed net 10% time savings even after editing. Other studies show 20-30% savings.

Concern 3: "I don't trust AI with my patients." Response: AI assists, but you remain in control. You review and approve everything before it becomes part of the medical record.

Concern 4: "My documentation style is too specialized." Evidence: UCLA study included 14 different specialties with positive results across all of them.

Financial Analysis: Is AI Scribe Worth the Cost?

Cost-Benefit Calculation

Typical AI Scribe Costs (2026):

Individual subscription: $100-200/month
Enterprise deployment: $150-300/physician/month
Implementation costs: $500-2,000 one-time

Time Value Calculation: Using UCLA study findings (41 seconds per note):

Scenario: Physician seeing 20 patients/day

Time saved: 13.7 minutes/day
Annual time saved: ~60 hours
Value of time (at $150/hour): $9,000/year
Annual tool cost: $1,200-2,400/year
Net benefit: $6,600-7,800/year

Additional Financial Benefits:

Billing optimization: 10-15% revenue increase from better documentation
Reduced burnout: Lower turnover costs
Increased patient volume: See more patients with saved time

ROI Timeline: Typically 3-6 months to positive return

Billing and Coding Impact

The UCLA study didn't specifically examine billing impacts, but clinical experience shows:

Documentation Completeness → Higher Billing Levels:

Better documentation supports higher E/M codes
AI prompts for missing elements needed for billing levels
Average billing increase: 10-20% where clinically appropriate

Example:

20 patients/day × 250 work days/year = 5,000 encounters/year
10% billing optimization × $150 average/encounter = $75,000 additional revenue
AI scribe cost: $2,400/year
Net billing benefit: $72,600/year

Specialty-Specific Considerations

Specialties Where AI Scribes Excel

Based on UCLA study and clinical feedback:

High-Value Specialties:

Primary Care: High visit volume, straightforward documentation
Psychiatry/Mental Health: Conversational encounters translate well
Pediatrics: Parent conversations captured effectively
Urgent Care: High volume, time-pressure benefits
Cardiology: Complex but structured assessments

Specialties with Unique Challenges

May Require Specialty-Specific Training:

Surgery: Operative notes have specific structure and terminology
Emergency Medicine: Rapid encounters, interruptions common
Radiology: Findings-based documentation, less conversational
Pathology: Microscopic findings, technical terminology

These specialties can still benefit but may need:

Specialty-specific templates
Additional training on verbalization of findings
Custom workflows for procedure documentation

The Future: Beyond the UCLA Study

Next Generation AI Scribe Features

Based on 2026 trends:

Emerging Capabilities:

Agentic AI: Beyond transcription to workflow management (learn more)
Multi-modal capture: Video, images, and audio integrated
Real-time guidance: AI suggests questions during encounter
Predictive documentation: AI anticipates likely documentation based on chief complaint

Coming Soon:

Autonomous order placement: AI generates orders for approval
Billing optimization: Real-time coding suggestions
Population health integration: AI links documentation to population metrics
Patient-facing summaries: Automatic after-visit summary generation

Regulatory Landscape

FDA Oversight:

AI scribes currently considered clinical decision support
Regulatory framework evolving in 2026
Some tools may require FDA clearance as AI capabilities expand

Reimbursement:

Medicare/Medicaid don't yet reimburse for AI scribe tools directly
Some commercial payers exploring coverage
Tax deductible as business expense

Conclusion: What Should You Do?

Action Steps Based on UCLA Study Evidence

If You're Considering AI Scribe Technology:

Key Takeaways from UCLA Study:

Time savings are real but modest: 10% reduction (41 seconds/note)
Burnout benefits are meaningful: Improved physician well-being
Accuracy requires vigilance: Always review AI-generated notes
Multi-specialty effectiveness: Works across diverse clinical settings
Safety events are rare: One mild event in 72,000 encounters

The Bottom Line:

The UCLA study provides high-quality evidence that AI scribe technology:

Reduces documentation time
Improves physician well-being
Works across multiple specialties
Is safe when used with appropriate oversight

For physicians struggling with documentation burden and burnout, AI scribes represent an evidence-based solution worth serious consideration.

Frequently Asked Questions

Medical Disclaimer: This content is for educational purposes only and should not replace professional medical judgment. Always consult current clinical guidelines and your institution's policies.