top of page

Schools Weigh Alternatives to Google Search

1 day ago
11 min read
Illustrative photograph. Photo: Firmbee.com / Pexels.

Google faces resistance from New York, Los Angeles, and Chicago schools over whether students can use Search without its generative AI features. The districts are considering Google Search alternatives, according to November 18 reporting from The Verge, citing The New York Times. They have not announced a completed migration.

The dispute concerns a distinction that matters beyond classrooms: disabling an AI application differs from removing AI features inside another service. Schools want to control the experiences students encounter. Google says AI is integrated into Search and points to changes intended to support learning.

That leaves administrators weighing familiar search access against their own instructional rules. DuckDuckGo illustrates an alternative approach, offering organizational settings that suppress AI-generated answers. Whether another provider meets a district’s broader requirements remains a separate question.

Google Search alternatives enter the district debate

The immediate development is a willingness to reconsider Search, rather than a decision to abandon Google’s entire education ecosystem.

The Verge’s account of the district search dispute identifies New York, Los Angeles, and Chicago as considering alternatives. It describes conditional responses, rather than signed replacement agreements or scheduled transitions. That distinction should shape expectations about what happens next.

According to that reporting, New York City schools chancellor Kamar H. Samuels said the district would explore other search options unless Google reconsidered. Chicago similarly linked consideration of another service to whether Google could provide Search without AI. Los Angeles was also identified as weighing alternatives.

The latest reporting builds on an earlier dispute. Google announced education product and control changes on November 3, 2026. The district responses reported on November 18 show how some large customers are evaluating the implications.

Those dates describe different developments. The November 3 announcement concerned Google’s products and planned controls. The November 18 story concerned schools’ responses. Neither establishes that students have already lost Search under the forthcoming arrangements.

The scope also requires care. A change to student search access does not establish that a district plans to replace its laptops, document software, or classroom platform. Treating those decisions as interchangeable would make the reported response sound broader than the available evidence supports.

A school can object to the behavior of its search engine while retaining other services from the same supplier. Conversely, changing the search destination does not settle every question about AI elsewhere. Administrators would still need to understand which services remain available and under what conditions.

That makes the districts’ conditional language meaningful. They are identifying a requirement and considering an alternative if the supplier cannot meet it. The unresolved issue is whether Google’s eventual controls satisfy that requirement.

There is also a difference between exploring options and demonstrating that an alternative works. Exploration can establish a negotiating position. A practical replacement requires evidence about search quality, student protections, accessibility, and administrative control.

The next stage therefore depends on specific decisions, rather than the intensity of public criticism. A district naming its requirements would clarify the problem. A published evaluation would clarify the available choices. An actual deployment would establish that consideration has become operational change.

For now, the story is about a familiar service becoming contestable because its behavior is changing. Schools are questioning whether their existing search arrangement still fits their instructional policies.

Schools want authority over the learning experience

The central pressure falls on administrators who must reconcile classroom rules with a supplier’s evolving product design.

Consider an illustrative classroom assignment that asks students to compare historical accounts. A teacher might expect students to identify authors, examine evidence, and explain disagreements. An automatically generated synthesis changes the starting point, even when it includes useful information.

That does not prove the synthesis harms learning. It does mean the teacher has a reason to decide whether it belongs in that particular activity. The instructional question concerns when assistance appears and what work the student must still perform.

Another assignment might deliberately involve AI-generated material. Students could examine an answer, verify its claims, and identify missing context. In that setting, access to the same feature could support the lesson.

These examples show why a general argument about whether AI belongs in education misses the immediate administrative problem. Schools need controls that match different activities, ages, and expectations. A product designed around a broad default can conflict with that need.

The search dispute makes that tension unusually visible. Teachers can assign a particular application for a particular task. Search, by contrast, often serves as the starting point for finding material across subjects.

A change at that starting point affects the conditions under which research begins. Students might encounter a generated explanation before opening a source. Teachers must then account for an interaction they did not necessarily assign.

Administrators also need policies they can explain. A statement that students may search but may not use generative AI becomes difficult to apply when both appear inside one service. The distinction must exist in the product experience, rather than only in school guidance.

Chicago’s response provides a concrete example. In November 12 coverage of Chicago’s AI restrictions, Chalkbeat reported that the district was exploring ways to suppress Google’s AI overview independently. It said Chicago would consider other search solutions if that could not be accomplished.

Chalkbeat also reported protests involving parents, educators, and students. Those reactions establish public concern, rather than a scientific finding about every AI feature. They help explain why administrators face pressure to give a clear answer.

District leaders must respond to families while maintaining usable research access. Teachers need workable expectations. Technology teams need settings that implement those expectations. Each group encounters a different part of the same decision.

The tradeoff is therefore practical as well as philosophical. Retaining a familiar service can simplify classroom routines. Enforcing instructional boundaries can require changing those routines. Neither objective disappears because the supplier describes its product as educational.

This is why the control question carries weight beyond a single chatbot. Schools are asking whether they can determine the terms under which students receive automated assistance.

An off switch does not guarantee AI-free search

Turning services off by default addresses automatic access, but it does not automatically provide the search experience schools are requesting.

Google’s position contains elements that should be evaluated separately. It plans to change how schools enable services. It is also developing an instructional approach for AI inside Search. Those measures address related concerns, but they perform different functions.

In its November 3 announcement about school AI controls, Google said relevant services would remain unavailable to student accounts until administrators enabled them. The company also said it would provide further implementation details before the changes.

An opt-in arrangement gives administrators an initial access decision. It does not, by itself, specify every feature included after a service becomes available. Schools still need to know what enabling Search permits students to encounter.

Google also described Guided Learning, an AI experience that uses questions and intermediate steps to help students work through problems. The company announced a Search pilot in selected U.S. schools during November, with broader expansion intended to follow.

Its examples included physics questions and book reports. Google said the experience would guide students through concepts rather than immediately supply a finished answer. That describes the company’s intended behavior, rather than independently verified performance across classroom situations.

The proposed approach recognizes a real instructional distinction. Offering a hint differs from completing an assignment. Asking a student to explain a step differs from producing the final response.

However, a school that wants no generative answers faces another question. A more instructional AI interaction still involves AI. Improving that interaction does not necessarily satisfy a policy requiring the feature to remain absent.

The confusion grows when the Gemini app and AI inside Search are discussed as one thing. Disabling the standalone application does not establish that generated search features disappear. Likewise, enabling Search does not necessarily mean enabling every separate Gemini product.

A useful account of Google school AI explained must preserve those distinctions. The relevant questions concern individual services, features within those services, and the settings that govern both.

For administrators, the decisive evidence will be the actual student experience after each setting changes. Product names and general assurances cannot establish whether a specific restriction works.

The distinction also applies to how Google’s response should be interpreted. Off-by-default access should not automatically be described as a reversal of its position on AI-free Search. That conclusion would require evidence that schools can enable Search while suppressing the disputed features.

Nor does the reporting establish that separating those features is technically impossible. Google’s integration argument explains its position. It does not independently settle what another product design could support.

The unresolved conflict remains narrow and consequential: Google is proposing control over access to services, while some schools want control over the experiences inside them.

Alternatives must satisfy more than an AI checkbox

A competing search provider gains an opening when its controls match school policy, but that opening does not establish suitability.

DuckDuckGo offers a useful comparison because its organizational documentation describes controls for disabling AI-generated answers. It also provides settings intended for environments such as schools. Those capabilities illustrate a different approach to separating search access from generated responses.

This comparison concerns the availability of controls, rather than a finding that DuckDuckGo delivers better results for every subject. It also does not establish that any district named in the reporting has selected the service.

A Google vs DuckDuckGo schools comparison therefore needs a defined scope. On the immediate question, the distinction concerns whether administrators can suppress generated answers while retaining search. A broader purchasing decision involves additional requirements.

Search relevance is one. Students need to locate appropriate material for assignments, including sources that support different reading levels and subjects. A control that removes an unwanted feature cannot compensate for results that fail the instructional task.

Accessibility is another. An alternative must work for the students and staff expected to use it. A district would need to examine the complete experience, rather than infer accessibility from the simplicity of a search box.

Administrative consistency also matters. A setting that works during an informal demonstration may behave differently across managed accounts or devices. Schools need evidence that the chosen configuration remains effective under their intended conditions.

The alternative must also fit existing safeguards. Disabling generated answers does not establish that every returned page is suitable for students. Search settings and broader content protections address different parts of the experience.

These are evaluation criteria, rather than reported defects in a particular provider. They explain why moving from public interest to an actual deployment requires more than choosing a familiar competing name.

There is a related limit to the phrase AI-free. A search engine can suppress generated answer features while still using machine learning elsewhere. The relevant policy question should identify the student-facing behavior a district intends to restrict.

Otherwise, schools and suppliers risk talking past each other. An administrator might mean that students should receive links without generated prose. A supplier might interpret the request as prohibiting every automated system involved in ranking results.

Clear requirements would make comparisons more useful. Schools could specify whether they need generated summaries removed, conversational interfaces disabled, or teacher-directed exceptions. Providers could then explain which requirements they support.

That would also make Google Search alternatives easier to evaluate without treating the dispute as a loyalty contest. The strongest replacement would be the service that meets the documented requirements in practice.

For competing suppliers, the opportunity lies in making those requirements verifiable. For schools, the challenge is determining whether a different search experience preserves research access while supporting the rules they have chosen.

Guided Learning needs evidence beyond its intended design

A feature designed to support learning should be judged by what students learn, alongside how it behaves during use.

Google’s instructional argument deserves examination on its own terms. If an AI system asks useful questions and supports practice, its value cannot be assessed solely by comparing it with an answer generator. Different designs can produce different experiences.

The reverse caution also applies. Calling an interaction guided does not establish that students perform the intended reasoning. A series of prompts can look instructional while still making the path to a finished answer too easy.

The important distinction concerns assistance during a task and learning that remains afterward. A student can complete an activity successfully with extensive help. That outcome does not automatically show independent understanding.

This creates a reasonable evaluation standard for Guided Learning. Schools should examine whether students can explain concepts, apply them elsewhere, and work without the same assistance. Those questions concern educational value rather than product engagement.

Independent research also argues against treating AI in education as a single intervention. Chalkbeat’s November 13 coverage of an education research review described a synthesis of 52 studies from Stanford’s AI Hub for Education.

The review found limited benefits across many student-facing uses, alongside potential value from carefully designed applications and support for teachers. Its authors emphasized thoughtful implementation and tracking whether tools benefit students.

Those findings do not establish that Google’s forthcoming Search experience causes academic harm. They also do not validate the specific pilot. The review covers varied interventions and settings, so conclusions about one implementation require appropriate evidence.

Chalkbeat’s earlier Guided Learning coverage presented another relevant concern. Common Sense Media’s Eisha Buch questioned whether schools had meaningful choice while emphasizing students’ need for AI literacy.

That position complicates a simple divide between supporters and opponents. A school can want students to understand AI while restricting its automatic presence during research. It can also approve a supervised activity while rejecting unrestricted access.

The same reasoning limits what can be concluded about alternatives. Removing generated answers does not guarantee that students evaluate sources carefully. Teachers still need assignments and assessment that reward that work.

For knowledge workers, the distinction has a familiar parallel. Collecting material, generating a summary, and understanding the evidence are different activities. A personal knowledge management approach can help frame those separate tasks without resolving the school procurement question.

The strongest skeptical position applies evenly to every proposal. Google should substantiate the benefits of its instructional design. Districts should evaluate whether restrictions achieve their educational goals. Alternative providers should demonstrate that their controls function as described.

That approach keeps the debate focused on observable outcomes. It avoids treating either automatic AI access or its removal as a complete learning strategy.

The next decisions will show whether choice is practical

The clearest signals will be usable controls, documented district decisions, and evidence from the Search pilot.

The first signal is Google’s implementation guidance ahead of the planned 2027 changes. Schools need a clear account of which services they can enable independently. They also need to know which AI features remain inside Search after access is restored.

That guidance will strengthen the districts’ concern if it leaves their requested restriction unsupported. It would weaken the concern if administrators can demonstrably preserve Search while removing the relevant generated features.

The distinction must be visible in ordinary use. A general commitment to flexibility is less informative than a documented setting and its resulting student experience. Clear explanations would also reduce confusion between disabling Gemini and changing Search.

The second signal is a district decision that moves beyond exploration. Published evaluation criteria would show what schools actually require. A named pilot, implementation plan, or selected provider would show that an alternative is becoming operational.

Such evidence would strengthen the case that dissatisfaction can translate into a change of service. Continued use of Google without a replacement plan would indicate that the reported consideration has not yet produced that outcome.

Neither result should be oversimplified. Remaining with Google might reflect acceptable revised controls, an unsatisfactory alternative, or an unfinished evaluation. A district’s explanation would matter as much as the decision itself.

The third signal is evidence from Guided Learning in Search. Feedback should address whether the intended experience works for real assignments. Evaluation should also distinguish successful assisted completion from understanding students can demonstrate independently.

Favorable evidence would support Google’s argument that its design addresses concerns about shortcuts. It would not settle whether every school should enable it. Weak results or easily bypassed boundaries would reinforce the need for caution.

Over the next few months, those signals should make the dispute easier to assess. Until then, claims of a completed school exodus exceed the reporting. Claims that new defaults have already resolved the disagreement also go beyond what has been established.

For other organizations buying AI-enabled software, the case offers a concrete question. Can a supplier’s controls implement the buyer’s policy, including when the buyer wants a particular feature absent? That question remains relevant even when the underlying service is familiar.

Schools considering Google Search alternatives should begin with the learning conditions they want to preserve, then ask suppliers to demonstrate the corresponding controls. Families can look for published explanations that distinguish planned changes from settings already available. Teachers can identify assignments where generated answers alter the work students must do. Technology teams can document whether proposed configurations hold up during ordinary use. These steps would turn a broad dispute about AI into a decision that people can inspect. The question to carry into 2027 is specific: can schools preserve useful search access while enforcing their chosen boundaries, and can the products involved demonstrate educational value under those conditions?

Get your work moving in remio

See what needs your attention and take the next step in Workspace.

remio currently supports Windows 10+ (x64) and Macs with Apple silicon.

Your AI Partner at Work
Get more done with remio

Plan. Create. Deliver.
All in one place.

bottom of page