AI Adoption Is Not Your Salvation

Not because AI is bad. Not because libraries shouldn\'t experiment with it. But because I recognize this pattern. I\'ve watched it before. We\'re treating the shiny new thing as the solution to problems it can\'t possibly solve.

I\'ve been doing library tech for 18 years. I\'ve watched us chase:

• Second Life islands (2007: "This is how we reach young people")
• Mobile apps that replicated our website badly (2010-2014)
• 3D printing makerspaces (2013: "Everyone will want this")
• Massive open online courses (2012: "This will replace schools")

Some of these helped. Some didn\'t. All of them distracted us from the actual crisis: we don\'t have enough staff, we can\'t afford to keep them, and when they burn out they take institutional knowledge with them. A robot cataloging machine doesn\'t fix that. Neither does AI.

Here's What the Research Actually Shows

The latest literature review on the future of U.S. librarianship synthesized research from 2019-2026 and came to a clear conclusion: AI adoption is accelerating, but its limitations are becoming clearer as real-world deployments multiply.

That\'s the diplomatic way of saying: "It works okay for some things, but it\'s not a game-changer."

The numbers:

• 67% of libraries exploring or implementing AI (up from 7% in early 2024)
• But accuracy on AI-generated subject headings? Low enough to require human review on every record
• That\'s not helpful. That\'s a liability.

Let me translate that for the non-tech librarians: If you're using AI to tag or catalog materials, 74% of the time it\'s wrong. That means either a human is fixing its mistakes (you just added a step), or patrons are getting bad results (you made search worse).

The Core Problem: AI Is a Copilot, Not a Pilot

The research conclusion is stark: "Human expertise remains essential. The "copilot" model is appropriate."

This matters. It means:

• You still need librarians
• You can't lay people off and replace them with AI
• Your competitor (a well-resourced library) isn't getting a magic cost-saving advantage
• But you might waste budget on tools that sound impressive in board meetings

What the successful deployments have in common: They augment human work, they don't replace it. A reference librarian using an AI search tool to suggest three starting points is powerful. An AI system trying to answer patron questions on its own is a disaster waiting to happen.

Where I Actually See AI Mattering in Libraries

I\'m not anti-AI. I\'m anti-hype. There are actually useful applications:

Need a reality check before you buy?

The research specifically recommends: "Prioritize patron-facing services (hotspots, streaming, digital literacy programs) over AI automation."

This is saying: Don\'t use your budget to automate your cataloging department. Use it to help patrons with digital literacy training about AI. Teach them how AI works, where it succeeds, where it lies. That\'s a real library service.

Staff Training, Not Panic

There\'s a significant readiness gap for AI literacy among library staff. Many of your team members don\'t understand how AI works. The solution isn\'t "more AI tools." It\'s staff training.

Your reference staff needs to understand:

• How LLMs hallucinate and confabulate (make things up convincingly)
• What training data likely went into popular AI tools
• When to recommend AI tools to patrons and when to warn against them
• How to verify AI outputs before trusting them

That\'s expertise. That\'s value. That's what we should be funding.

The Real Crisis You Should Be Worrying About

Here's what I actually want to talk about:

Public library job satisfaction has declined significantly, according to recent industry surveys.

Your best reference librarian just quit. Your youth services director took a job in corporate training. Your systems administrator is burnt out and won't return emails for three days. You have 15% more work and 20% fewer people to do it.

That\'s not a problem AI solves. That\'s a problem management creates by saying "Let\'s try this AI thing" instead of "Let\'s pay people what they're worth."

The research is explicit about this: "Urgent retention focus is required - competitive compensation is necessary but insufficient without addressing toxic culture, recognition deficits, and book challenge stress."

Your staff is burning out because:

• They're underpaid for their expertise
• They're being personally attacked by people trying to ban books
• They're being asked to do more with less
• They\'re told the solution is "use AI better" instead of "we\'re hiring you help"

AI doesn't fix any of that. Budget spent on AI tools instead of staff salaries makes it worse.

The Hard Conversation Your Board Needs to Have

When your director or consultant pitches "AI implementation," ask these questions:

1. Is this solving a problem or creating a project? Some AI implementations are just homework assignments disguised as strategic planning. "Let\'s implement AI" isn\'t a strategy. "We're using AI to reduce the time librarians spend on X task by Y% so they can focus on Y" is a strategy.

2. What\'s the accuracy threshold for acceptance? If the AI tool still requires human review on every record, why are we using it? If it\'s 92% accurate at something, that's different. Define success upfront.

3. Does this require staff hours to fix? If the answer is yes, you haven\'t saved money. You\'ve just moved the work from machines to humans. If you don't have staff now, adding "fix AI outputs" to their workload is not a solution.

4. Have we tried paying people better first? Seriously. Before implementing new tech, do the basics. Ask staff what\'s broken. They\'ll tell you. Usually it\'s "I can\'t afford rent" or "I\'m getting death threats because of book challenges" or "I haven\'t had a real break in three years."

What You Should Actually Be Investing In

The research gives clear guidance on resource allocation:

Retention Over Recruitment

50% of state library leadership is retiring. That exodus is happening right now. Succession planning is critical. You can't recruit your way out of this. You have to keep the people you have.

Money spent on competitive compensation, professional development, and actual support beats any AI implementation.

Diversity Initiatives in MLIS Programs Are Showing Results

The future workforce is coming from schools that are intentionally diversifying. Support that. It matters for representation. It matters for cultural competence. It matters because the communities you serve deserve staff that reflects them.

Social Services Infrastructure

55+ libraries now have full-time social workers. That\'s not a trend. That\'s the future. Libraries aren\'t book warehouses anymore. They\'re community anchors. Staff that can actually help people navigate housing applications, mental health crises, and workforce development are worth far more than AI chatbots.

Community Partnerships Over Internal Capability Building

You don't need to build everything in-house. Partner with social services agencies, workforce development boards, healthcare providers. Their expertise is deeper. Your budget is better spent paying for those partnerships than trying to hire and train staff for things outside your core mission.

The Uncomfortable Truth

AI isn\'t your salvation because your problems aren\'t technical. They're structural.

You're underfunded. Your staff is burnt out. Your communities are facing attacks on intellectual freedom. Patrons need actual human services, not automated ones.

A chatbot won\'t fix funding crises. A subject heading generator that still requires human review on every record doesn\'t make your reference librarian less critical. An AI-powered search tool won\'t help when the person asking can\'t afford housing.

Yes, use AI. For the things it actually helps with. For training your staff about it. For showing patrons what it can and can't do.

But spend your real money on people. On salaries. On retention. On support for staff dealing with censorship battles and burnout.

That\'s not a technology problem. That\'s a values problem.