From the Transportation edition of October 29, 2026
Giving Boston the Green Light: lessons from AI signal-timing recommendations
Eno Center for Transportation (author: former Boston Chief of Streets) · City traffic signals · Boston, Massachusetts, US
- Publisher
- Eno Center for Transportation
- Original publication
- August 20, 2026
- Source retrieved
- 2026-10-30
What happened
Boston implemented Google Project Green Light recommendations at more than a quarter of its 900 signalized intersections, averaging a 13% delay reduction at modified intersections (27% at several Back Bay intersections), but reversed changes at about a third of locations for lack of improvement or negative impacts. The AI sometimes misread phasing at complex intersections, does no corridor or network analysis, surfaced equipment failures rather than new insights at some sites, and cannot resolve pedestrian-versus-vehicle value tradeoffs.
Why it matters
Why now: state DOTs such as Caltrans have adaptive AI signal control in proof of concept this month; Boston's account shows what advisory, human-reviewed signal AI does and does not deliver.
Evidence and measured results
First-person practitioner account with city-reported averages; no published methodology or independent verification of the 13% figure.
Limitations and uncertainty
Author is a former city official; impact measurement is vendor-provided; localized optimizations may not yield network benefit.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-10-29; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Cities cannot retime hundreds of signals often enough with existing staff and tools.
- Stakeholders
- City signal engineering
- Transportation department leadership
- Pedestrian and bike program staff
- Discovery questions
- How many intersections are retimed per year today?
- Who reviews and approves timing changes?
- How do you measure impact after a change?
- Value hypothesis
- Advisory AI with fast feedback could raise retiming throughput with modest average delay reductions.
- Potential engagement
- A bounded measurement plan for advisory signal recommendations on a defined set of intersections.
- Unsupported claims
- Corridor or network throughput gains
- Safety improvements
- A guaranteed 13% delay reduction
Pre-sales engineering
Role takeaway
- Fit
- Fits cities with limited integration capacity; advisory only.
- Architecture and integration
- Phone movement data, inferred phasing, a recommendations dashboard, manual implementation and post-change reporting.
- Prerequisites
- Accurate timing plan records
- Working detectors
- Engineer review capacity
- Constraints
- No corridor or network analysis
- Pedestrian crossing minimums
- Security
- Low control-system exposure since there is no direct integration; confirm data-sharing terms for movement data.
- Proof of value
- Independent before/after delay measurement
- Rollback rate tracking
- Phasing-inference accuracy at complex intersections
Delivery
Role takeaway
- Work
- Run a recommend, approve, implement, measure and roll-back cycle across selected intersections.
- Dependencies
- Signal maintenance crews
- Vendor recommendations
- Timing plan documentation
- Ownership
- City signal engineering.
- Skills and adoption
- Engineers learn to evaluate AI-inferred phasing and impact reports.
- Governance checkpoints
- Engineer approval per change
- Post-change review within weeks
- Pedestrian standard check
- Acceptance criteria
- Proposed: documented delay change per modified intersection
- Proposed: rollback decision recorded within an agreed window
- Risks
- Localized gains shifting queues elsewhere
- Phasing inference errors
- Over-reliance on vendor-reported impact
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Advisory-only design with no integration into signal control let Boston deploy in weeks; built-in post-change reporting provided the validation loop.
Governance
Who approves, reviews and stays accountable for outcomes?
Engineers approve every change; values-based tradeoffs such as crossing-time minimums and bike signals stay with people.
Security and privacy
What data, permissions and controls need testing?
Relies on anonymized phone movement data; no direct control-system integration limits cyber exposure.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Pedestrian standards (at least 20 seconds for a 45-foot crosswalk) constrain optimization; engineer review time is required.
Procurement
What should contracts, pricing and exit terms secure?
Low-integration advisory tools are faster to adopt than closed-loop adaptive systems.
Operating model
Which teams own the service once it runs?
A short feedback loop (weeks) between change and measured impact, with rollback as a normal outcome.
Publication history
- 2026-10-29Transportation · Issue 0112 resources
Stable resource ID: boston-project-green-light-signal-lessons