A city council meeting can determine the future of a neighborhood.
A housing proposal can affect rent and development.
A transportation decision can change a daily commute.
A zoning vote can reshape a business district.
Yet residents cannot meaningfully participate in those decisions if they cannot understand what is being discussed.
That is the problem Mountain View, California, is now trying to address with an experiment that could have implications far beyond Silicon Valley.
The city has begun piloting an artificial-intelligence translation service called Wordly, allowing people attending selected public meetings to follow discussions through real-time translated captions and audio in more than 60 languages. The pilot is currently being used for City Council, Environmental Planning Commission and Rental Housing Committee meetings.
For Mountain View, the technology represents an attempt to expand language access without requiring a human interpreter for every language at every meeting.
But it also raises difficult questions.
Can AI interpret the language of government accurately enough for consequential public discussions?
What happens when a translation gets an important legal, planning or financial term wrong?
Could automated translation genuinely increase civic participation, or will it simply create another digital barrier?
And where should the line be drawn between AI assistance and human interpretation?
Mountain View’s pilot is becoming a useful real-world test of those questions.
Why Mountain View Is Testing AI Translation
Mountain View is not a small or linguistically uniform community.
Recent U.S. Census data shows that about 42.9% of the city’s residents are foreign-born, while 49.5% of residents age five and older speak a language other than English at home. The city’s population is also racially diverse, with Asian residents representing 36.3% of the population and Hispanic or Latino residents representing 17.5%.
Those numbers help explain why language access is not a minor administrative issue.
For a significant share of residents, English is not the language most often used at home.
A city government that communicates only in English may therefore be leaving a large portion of its residents at a disadvantage when decisions are made publicly.
Mountain View already provided some language support before the AI pilot.
Spanish and Chinese interpretation services were available for City Council meetings when requested in advance.
The AI pilot expands the potential language menu dramatically.
From Two Languages to More Than 60
Previously, Mountain View’s in-person interpretation offering for City Council meetings was focused on Spanish and Chinese.
The Wordly system can provide translated captions and audio in more than 60 languages.
That difference changes the economics of accessibility.
Providing a professional interpreter for one language can require advance scheduling, staffing and specialized equipment.
Providing human interpretation for dozens of languages at the same meeting would be far more difficult and expensive.
AI translation does not eliminate those costs entirely, but it can provide a much broader baseline service.
A resident who speaks Vietnamese, Korean, Tagalog, Japanese, Portuguese, Arabic or another supported language can potentially access the same meeting without requiring the city to find a separate interpreter for that language in advance.
How the System Works
The system is designed to be relatively simple for attendees.
During a participating meeting, users can scan a QR code and select their preferred language.
They can then follow the meeting by reading translated captions on their own device, listening to translated audio through headphones, or doing both. The service can also be accessed remotely.
That matters because accessibility depends heavily on convenience.
A complicated system can discourage people before they even begin.
The Mountain View approach essentially turns a smartphone into a personal interpretation device.
Instead of placing one interpreter in a booth translating for a room full of people, the system allows different attendees to choose different languages simultaneously.
One person could listen in Spanish.
Another could read Chinese captions.
Another could follow Vietnamese.
Another could remain with English.
All could participate in the same meeting.
The Pilot Began Before the City Council Trial
The AI service first appeared at an Environmental Planning Commission meeting before being used at a Mountain View City Council meeting on September 8, 2026.
That sequence is significant.
Testing the system on an advisory or commission meeting before deploying it during a major council session gives city staff an opportunity to identify technical problems.
Government meetings have specialized vocabulary.
Planning discussions may involve zoning classifications, environmental review, development standards and technical terminology.
Rental housing discussions can involve legal terms, regulations and financial language.
A translation system needs to handle far more than casual conversation.
A Real-World Test Came From a Chinese-Speaking Resident
One of the earliest public experiences with the system came from Mountain View resident Tootoo Thomson.
After scanning the QR code and selecting Chinese, Thomson followed the meeting on her phone using translated captions.
She described the experience as about 98% accurate, while noting that some phrases were translated awkwardly from English into Chinese. She also appreciated the ability to translate comments from people speaking different languages in real time.
That anecdote captures both the promise and the limitation of AI translation.
For ordinary conversation, an imperfect phrase may be harmless.
At a city meeting, however, context matters.
A slightly awkward sentence can be amusing.
A mistranslated policy requirement can be consequential.
Why Government Language Is Especially Difficult
Government is full of specialized language.
Officials may use terms that have specific meanings within a legal or regulatory framework.
Consider phrases such as:
Affordable housing requirement.
General plan amendment.
Conditional use permit.
Environmental impact.
Rent stabilization.
Public right-of-way.
Capital improvement.
Development agreement.
These phrases cannot always be translated successfully by treating each word separately.
The system must understand the context.
That is one reason AI translation providers recommend custom glossaries containing organization-specific names, technical terms and acronyms. Wordly says such glossaries can improve transcription and translation accuracy, while also warning that errors remain possible.
For a city government, building and maintaining those terminology libraries could become an important part of the pilot.
Accuracy Is Not the Same as Reliability
AI translation companies often report very high accuracy rates under controlled conditions.
Wordly says its translation accuracy can reach around 98% to 99%, depending on language pair and audio conditions.
But that should not be interpreted as a guarantee that 98% of every city-council sentence will be correct.
Translation quality depends on several factors:
Audio quality.
Speaker accents.
Background noise.
Speaking speed.
People talking over one another.
Technical vocabulary.
Language pair.
Internet connectivity.
Microphone placement.
Context.
A meeting room is not a laboratory.
A person speaking clearly into a microphone is much easier for an AI system to process than several people interrupting each other.
The Delay Matters Too
Real-time translation is not actually simultaneous in the strictest sense.
The software processes speech, determines context and produces the translated output.
Wordly says translated audio is generally delivered after a sentence has been completed, creating a slight delay.
For passive listening, that may be acceptable.
But it can become more complicated during rapid exchanges.
Imagine a public hearing where a resident speaks, a council member responds and another official immediately asks a follow-up question.
A short delay can make it harder to track the conversational flow.
The system therefore needs to be judged not only by sentence-level accuracy but also by whether users can follow the meeting comfortably.
Captions May Be More Important Than Audio
One of the advantages of the system is that users do not have to listen to synthetic translated speech.
They can read the translated captions on their phones.
That can be particularly valuable when audio translation is slightly delayed or when users want to compare the original speech with the translated version.
Captions also provide an accessibility benefit for people who are deaf or hard of hearing, although the system is primarily being introduced for language access. Wordly says its live captions can support users who prefer reading rather than listening.
AI Translation Does Not Replace Human Interpreters
This distinction is central to Mountain View’s approach.
The city has explicitly said that its AI service is a supplement rather than a replacement for existing in-person Spanish and Chinese interpretation during City Council meetings.
Residents who need human interpretation can still request it in advance.
Requests must be submitted by 5 p.m. two business days before the meeting.
That creates a two-level system.
AI provides broad, flexible access.
Human interpretation remains available when greater assurance is needed.
This may become an important model for other cities.
Why Human Interpreters Still Matter
Professional interpreters do much more than convert words between languages.
They can recognize ambiguity.
They can clarify context.
They can manage turn-taking.
They can understand cultural nuances.
They can ask for clarification.
They can adapt to unexpected situations.
AI systems cannot necessarily perform all those functions reliably.
That matters particularly when meetings involve legal rights, public benefits, housing disputes, emergency decisions or other high-consequence topics.
The safest model may therefore be “AI first for broad access, human interpretation where precision is especially important.”
The Pilot Has a Relatively Modest Price Tag
According to Mountain View officials, the total cost of the pilot is about $58,367.
That includes a $43,750 contract with Wordly for 300 hours of live translation access, with the remaining funds used to make the Council Chamber technologically capable of supporting the service.
The city’s adopted budget had previously included about $60,000 in limited-period funding for an AI translation initiative.
That makes the pilot relatively inexpensive compared with the potential cost of staffing dozens of human interpreters across numerous meetings.
But cost should not be the only metric.
The real question is cost per meaningful participant served.
What the City Actually Wants to Measure
Mountain View says it plans to collect feedback from residents who use the system and evaluate translation accuracy and the program’s real-world impact on language access.
That is important because deployment alone does not prove success.
The city could technically provide translation in 60 languages while barely increasing public participation.
Alternatively, a relatively small number of users might gain substantial access to government information.
Those outcomes require measurement.
Measuring Civic Participation Could Be the Most Important Test
The ultimate test should not be:
Did the AI translate the meeting?
It should be:
Did more people participate because the AI translated the meeting?
That could be measured in several ways.
How many residents use the tool?
Which languages are most frequently selected?
Do users stay for longer portions of meetings?
Do more multilingual residents submit public comments?
Does attendance increase among residents who previously participated less?
Do users report greater understanding of city decisions?
Do people use the system remotely as well as in person?
Those data points would tell the city whether language technology is changing civic engagement rather than merely improving convenience.
Public Comment Is Where the Stakes Rise
Watching a translated meeting is one thing.
Speaking at a public meeting is another.
Mountain View’s budget planning documents had envisioned residents being able to use AI translation not only to follow meetings but also to interact and potentially participate in public comment.
That is a much more ambitious goal.
For a resident to participate meaningfully, they need to understand the proceedings and communicate their own position.
This creates additional technical challenges.
The system has to recognize the speaker’s language.
It has to transcribe accurately.
It has to translate the comment.
And everyone listening needs to receive the translated output quickly enough to understand the exchange.
That is a much harder task than simply translating a prepared speech.
Public Meetings Are Not Just Information Sessions
A common misunderstanding is that translation is mainly about helping residents understand what government is doing.
That is only half of the issue.
Public meetings are also mechanisms for democratic participation.
Residents can challenge decisions.
Ask questions.
Share personal experiences.
Oppose proposals.
Support policies.
Request changes.
Communicate community needs.
Language access therefore affects the ability to exercise civic voice.
That gives the Mountain View pilot significance beyond technology.
Housing Decisions Make the Issue Especially Relevant
Mountain View’s Rental Housing Committee is one of the bodies included in the pilot.
That is important because housing is an area where language barriers can have direct consequences for residents.
A discussion about rental policy may affect tenants, landlords, developers and housing advocates.
Not every resident who is directly affected will be equally comfortable discussing the issue in English.
A translation system could make it easier for more people to understand changes that affect their homes.
Planning Decisions Can Also Have Long-Term Effects
The Environmental Planning Commission is another body using the pilot.
Planning meetings may discuss projects that remain in the city’s landscape for decades.
A new apartment development.
A transportation project.
A zoning change.
A commercial redevelopment.
An environmental review.
These decisions can affect people who never attend a meeting unless the process feels accessible.
Language access can therefore influence who participates in shaping the physical city.
Mountain View’s Diversity Reflects Silicon Valley
The experiment is particularly relevant in Silicon Valley because the region combines global migration, technology companies and highly educated international communities.
People may work in technology and finance while still preferring another language at home.
That creates an interesting contradiction.
A person may be comfortable using English professionally but prefer to understand complex local government decisions in another language.
Language preference is not necessarily a measure of education.
It is often a measure of context.
A resident may be perfectly capable of understanding English while still finding legal or policy language easier to process in the language they use most at home.
Digital Access Is Another Barrier
AI translation does not eliminate every accessibility problem.
The Mountain View service requires users to interact with a digital interface through a personal device.
That assumes people have:
A smartphone or suitable device.
Battery power.
Internet connectivity.
Enough technological familiarity to scan a QR code.
The ability to navigate the interface.
For many residents these requirements are trivial.
For others, they are barriers.
That is why hands-on assistance from city staff during some meetings is important. Mountain View says it plans outreach and practical support to help residents learn the system.
The QR Code Is Simple—But It Changes the Experience
The QR-code approach has an interesting advantage.
Users do not necessarily need to install a dedicated application.
They can scan the code and enter the translation session using their device.
This lowers the entry barrier compared with systems that require accounts, downloads or specialized equipment.
But the city will still need to ensure that QR codes are clearly displayed and that first-time users understand what to do.
Technology is only accessible when people know how to use it.
Remote Participation Could Be One of the Biggest Benefits
Mountain View’s system can be accessed remotely as well as inside Council Chambers.
That could prove particularly important for residents who cannot attend city meetings in person.
Parents may have childcare responsibilities.
Workers may not be able to leave their jobs.
Residents with mobility limitations may find travel difficult.
People may simply prefer to participate from home.
Adding multilingual translation to remote participation makes civic access more flexible.
It also means the pilot could eventually influence how cities approach online public meetings more broadly.
The Technology Could Produce a Multilingual Public Record
Mountain View officials say the public can access full meeting transcripts and summaries in the languages Wordly offers even after meetings conclude.
That could become one of the most significant long-term benefits if implemented consistently.
A multilingual archive would allow residents to review discussions later rather than having to understand everything in real time.
It could also help community organizations track issues over time.
However, the quality question becomes even more important when translated text remains available as a public record.
A temporary translation error during a live meeting is one problem.
A wrong translation preserved online indefinitely is another.
AI Translation Could Make Government More Searchable
There is also a future possibility that goes beyond basic translation.
Imagine being able to search a city’s meeting archive for a concept in your preferred language and immediately locate every discussion about:
Rent stabilization.
Bike lanes.
Water rates.
School boundaries.
Homelessness.
Development projects.
Public safety.
That could transform how residents interact with local government.
Instead of attending dozens of meetings just to find one discussion, they could search a multilingual archive.
This would turn translation from an accessibility tool into a discovery tool.
But Transcripts Raise Privacy Questions
Any system that processes public-meeting audio and generates transcripts raises questions about data handling.
Who stores the audio?
How long is it retained?
Is the transcript used to improve the AI model?
What information is shared with the vendor?
Are personal details included?
Does the system process every speaker equally?
These questions become more important as governments move AI from experimental projects into routine operations.
Mountain View’s pilot should therefore be evaluated not only for accuracy and usability but also for data governance.
The City Is Developing a Broader AI Policy
The translation pilot is part of a larger conversation inside Mountain View about responsible AI use.
The city’s budget documents say staff are developing an AI Acceptable Use Policy while exploring how AI could support the Council’s strategic priority of “Community for All.”
That connection matters.
AI adoption in government is not simply about finding useful software.
It also requires deciding:
What uses are acceptable?
What oversight is required?
Which decisions can be automated?
Where must humans remain responsible?
How should residents be informed?
How should errors be corrected?
Translation may be one of the relatively lower-risk places for a city to experiment with AI, but the lessons learned could influence other applications.
Could AI Translation Become a Standard Government Utility?
That is increasingly plausible.
The same technology could eventually support:
City council meetings.
Planning hearings.
School board meetings.
County meetings.
Public health briefings.
Emergency information sessions.
Community workshops.
Public consultations.
Transportation hearings.
Regional planning meetings.
Instead of scheduling language services from scratch every time, governments could have translation available as part of their standard meeting infrastructure.
The Economics Become More Attractive at Scale
AI translation becomes particularly interesting for organizations holding many multilingual meetings.
A human-interpreter model requires arranging people for each language and event.
An AI platform can potentially be reused across hundreds of meetings.
That does not necessarily mean AI is cheaper in every situation.
But the marginal cost of adding another supported language may be substantially lower.
This creates a compelling economic argument for public-sector experimentation.
Yet Governments Should Resist the “Cheap Interpreter” Mindset
There is a danger in presenting AI simply as a cheaper replacement for professional interpreters.
The more productive way to think about it is as a new layer of accessibility.
Human interpreters remain essential for situations where accuracy, nuance or legal consequences are particularly important.
AI can fill gaps.
It can serve more languages.
It can support last-minute needs.
It can provide captions.
It can help people follow meetings remotely.
Those advantages are meaningful even without eliminating human interpretation.
Language Access Can Strengthen Trust in Government
Trust is often influenced by whether people feel heard.
A resident who repeatedly encounters government information only in a language they find difficult may conclude that government is not designed for them.
Providing a way to follow proceedings in their preferred language sends a different message.
It says:
Your participation matters.
Your language is recognized.
You have a place in this process.
That symbolic effect may be as important as the translation itself.
But Trust Can Be Destroyed by a Major Error
The reverse is also true.
Imagine a resident depends on an AI translation and misunderstands a deadline, legal requirement or policy proposal because the translation is wrong.
Confidence could decline rapidly.
That is why transparent disclaimers are important.
Mountain View explicitly warns users that it cannot guarantee the accuracy of AI-translated text or interpretation.
That warning is not a weakness of the program.
It is an acknowledgment that automated translation should not be treated as infallible.
What Could Make the Pilot More Successful?
The city can improve the system in several practical ways.
First, it can build comprehensive glossaries containing Mountain View-specific terminology, names, agencies, neighborhoods and recurring policy language.
Second, it can improve microphone and room-audio quality.
Third, meeting participants can be encouraged to speak clearly and avoid talking over each other.
Fourth, staff can provide assistance to first-time users.
Fifth, the city can publish clear guidance explaining when residents should use AI translation and when they should request human interpretation.
These recommendations align with Wordly’s own best-practice guidance, which highlights audio quality, clear speech, stable connectivity, one speaker at a time and custom glossaries as important factors in translation quality.
The Pilot Could Also Reveal Which Languages Are Most Needed
One unexpected benefit could be better data about language demand.
The city can potentially see which languages residents select most often.
That information could inform future investment.
Suppose Spanish and Chinese remain dominant.
The city might continue emphasizing human interpretation for those languages.
Suppose Vietnamese, Korean or Tagalog demand grows.
Officials might prioritize translated materials or community outreach in those languages.
Technology can therefore help government understand its own community.
Translation Is Only One Part of Accessibility
The larger challenge is broader than language.
Residents also need:
Clear agendas.
Plain-language explanations.
Accessible websites.
Translated documents.
Affordable transportation.
Remote participation.
Reasonable meeting times.
Disability accommodations.
Childcare where appropriate.
Accessible public-comment procedures.
A resident may understand every word of a meeting and still be unable to participate because the process is too difficult to navigate.
AI translation should therefore be viewed as one component of a larger accessibility strategy.
Plain English Still Matters
Interestingly, AI translation does not remove the need for simpler government language.
The clearer an official speaks, the easier it is for both humans and machines to understand.
That means governments may discover a secondary benefit from AI translation:
It could encourage officials to speak more clearly.
Shorter sentences.
Fewer unnecessary acronyms.
Less jargon.
More direct explanations.
These improvements help everyone, including native English speakers.
A Better Public Meeting Could Be More Multilingual Than Any Previous One
Imagine a future Mountain View meeting where dozens of residents follow the discussion in their preferred languages.
One resident reads Spanish captions.
Another listens in Mandarin.
Another follows Korean audio.
Another uses English.
A resident participating from home follows the same discussion.
After the meeting, everyone can access translated summaries.
That would represent a significant transformation in local democracy.
The meeting would no longer be linguistically centered on a single language.
The Broader Silicon Valley Experiment
Mountain View is a particularly interesting place to test this because technology is embedded in the city’s identity.
The community is surrounded by companies developing artificial intelligence, cloud computing, speech recognition and machine translation.
Yet the city is testing whether those technologies can solve an ordinary civic problem.
That creates a fascinating inversion.
Silicon Valley technology is often associated with ambitious commercial products.
Mountain View is asking whether it can help residents understand a zoning meeting.
That is a much more grounded measure of technological usefulness.
Could Other Cities Follow?
Very likely.
Cities across California and the United States have increasingly diverse populations.
Many governments already struggle to provide interpretation in every language their residents speak.
AI could offer a practical way to expand basic coverage.
Smaller cities may find the technology especially attractive because they cannot afford large interpretation teams.
Larger cities could use it as an additional service alongside professional interpreters.
The Biggest Question Is Not Whether AI Can Translate
AI can translate.
The harder question is whether it can translate well enough for democracy.
That standard is different from translating a restaurant menu or a casual conversation.
A public meeting contains policy, emotion, technical information, personal testimony and legal consequences.
Accuracy matters.
Timing matters.
Context matters.
Trust matters.
The standard should therefore be higher than simply producing a grammatically understandable sentence.
What Mountain View Should Measure
A serious evaluation should examine at least four categories.
Accuracy: Are important terms and meanings preserved?
Accessibility: Can residents actually use the system without difficulty?
Participation: Does multilingual civic engagement increase?
Trust: Do residents feel more confident that they understand the proceedings?
Cost should be measured too, but it should not become the primary success criterion.
The cheapest system is not necessarily the best system.
The best system is the one that expands meaningful participation responsibly.
A Model of AI and Human Collaboration
The strongest outcome for Mountain View may not be choosing between AI and humans.
It may be proving that both can coexist.
AI handles breadth.
Humans handle nuance.
AI makes last-minute access possible.
Humans provide high-assurance interpretation.
AI provides captions.
Humans intervene when precision matters.
AI creates a broader baseline.
Human professionals handle the highest-stakes situations.
That division of labor could become a model for government agencies experimenting with artificial intelligence.
Final Takeaway
Mountain View’s AI translation pilot may look like a small technology project, but it is testing a much larger idea: Can artificial intelligence make local democracy more accessible without sacrificing accuracy or trust?
The city’s demographics make the question particularly relevant. Nearly half of Mountain View residents age five and older speak a language other than English at home, and the foreign-born population is substantial.
The Wordly pilot now gives residents access to real-time translated captions and audio in more than 60 languages during City Council, Environmental Planning Commission and Rental Housing Committee meetings.
The technology is inexpensive compared with providing human interpretation in dozens of languages, and early user feedback suggests that it can be remarkably effective while still producing occasional awkward or inaccurate translations.
But that imperfection is precisely why the experiment matters.
Government cannot simply ask whether a machine produces a translation.
It has to ask whether a resident can understand a decision that affects their life.
It has to ask whether someone can follow a housing discussion, challenge a development proposal or submit public comment without being disadvantaged by language.
And it has to ask whether residents trust the system enough to use it.
Mountain View has wisely kept human Spanish and Chinese interpretation available rather than treating AI as a complete replacement.
That hybrid approach may ultimately prove to be the most practical future.
AI can make multilingual access dramatically broader.
Professional interpreters can provide a safeguard when accuracy and nuance matter most.
Together, they could help transform public meetings from events that technically welcome everyone into meetings where more residents can actually understand, participate and be heard.
That is a much more meaningful test of artificial intelligence than whether a chatbot can write a poem.
It is a test of whether technology can help make local government work better for the people it serves.
5 Frequently Asked Questions
1. What is Mountain View’s AI translation pilot?
Mountain View is testing Wordly, an AI-powered service that provides real-time translated captions and audio during selected public meetings. The pilot currently covers City Council, Environmental Planning Commission and Rental Housing Committee meetings.
2. How many languages can the AI service translate?
The city says Wordly provides access to more than 60 languages. Residents can select their preferred language and use their own device to read translated captions, listen to translated audio, or both.
3. Does AI translation replace human interpreters in Mountain View?
No. The city continues to provide in-person Spanish and Chinese interpretation for City Council meetings upon advance request. The AI system is being tested as an additional language-access option rather than a complete replacement for professional interpreters.
4. How accurate is AI translation during public meetings?
Accuracy varies according to the language pair, audio quality, speaking style, terminology and other conditions. Wordly says its translation accuracy can reach roughly 98% to 99% in appropriate conditions, while Mountain View itself warns that it cannot guarantee the accuracy of AI-generated translations.
5. Why is Mountain View testing AI translation?
The city wants to determine whether AI can improve language accessibility and help its diverse community participate more easily in local government. The pilot is also intended to provide evidence about real-world accuracy, usability and civic impact before the city decides how the technology should be expanded or adapted.
Sources San Jose Spotlight

