As employers rush to adopt AI-powered technology, many HR professionals are being asked to move quickly without fully understanding how the technology works, how risks are evaluated, or who is accountable when something goes wrong.
Vendor demonstrations often emphasize speed, efficiency, and productivity while giving less attention to more difficult questions involving bias, transparency, data privacy, accountability, and regulatory compliance. As scrutiny of workplace AI increases, HR cannot afford to rely on a vendor’s assurances or defend a hiring decision by saying, “The AI recommended it.”
The challenge is particularly significant for HR, which is increasingly responsible for managing sophisticated technology despite often lacking deep technical expertise.
“HR is by default one of the least technical departments,” said Mike Peditto, fractional recruiting leader, HR tech advisor, and content creator, speaking at the HR Tech 2026 conference in Las Vegas.
Peditto has observed HR professionals navigating vendor demonstrations and technology sales pitches, and said he frequently sees buyers overwhelmed by presentations without asking the questions that matter.
“I see people at the expo hall being spun around by sales reps,” he said. “People sitting through demos, not knowing what’s going on, just nodding.”
He said it is essential for HR pros to become more involved in technology purchasing decisions, advance their understanding of what vendors are selling, and establish clear expectations for how tools will be used.
His advice is to approach AI purchasing through three areas of consideration: understanding what the organization needs; understanding what vendors are offering; and understanding how the organization intends to implement the technology.
Here are 10 questions HR practitioners should ask before purchasing an AI tool.
Diagnose Your Organization's Needs
The evaluation process should begin before HR schedules a vendor demonstration or starts shopping for new software, Peditto said. Leaders first need to understand their own processes, existing technology, and decision-making responsibilities.
1. What specific problems are we trying to solve, and how do we know they exist?
AI adoption should begin with a clearly defined business problem, not an executive mandate to acquire the newest technology. Without understanding the underlying need, organizations risk purchasing expensive tools that don’t address their actual challenges.
“What you need AI for is different than what another company needs it for,” Peditto said.
HR leaders should identify the specific outcomes they expect a tool to deliver. Are recruiters spending too much time screening applications? Are hiring managers struggling to identify qualified candidates? Is the organization having difficulty measuring skills or predicting workforce needs?
The answers should determine whether AI is appropriate. “Maybe instead of buying new AI, your process just sucks,” Peditto said. “Maybe you don’t need a new expensive piece of software.”
2. What technology are we already using, and are we using it correctly?
Before purchasing another platform, HR should take inventory of its existing technology.
Peditto explained that organizations frequently pay for software that employees barely use or that already includes capabilities they are considering purchasing elsewhere.
“Now if you have a piece of technology and you hate it, that’s different,” he said. “But before you go out shopping for new technology, figure out if you already have it.”
This review should examine user adoption, available features, licensing costs, and whether existing tools can accomplish the desired objectives.
3. Do employees know how to use the tools we already have?
Technology investments don’t automatically translate into better performance. Employees need training and support to use software effectively.
Peditto described organizations that purchase technology, fail to adequately train employees, and eventually abandon the platform in favor of another product.
He recalled hearing about a company that purchased software from a well-known vendor but found that nobody knew how to use it. The vendor had provided little support after the sale, prompting the organization to purchase another solution. The result was wasted money and another implementation challenge.
“Put some money into training and enablement,” he said.
4. Who owns the purchasing decision, and who needs to be involved?
HR technology purchases often involve multiple stakeholders, including HR, IT, legal, procurement, information security, and senior executives.
Before beginning the vendor evaluation, organizations should establish who has final decision-making authority, who provides input, and who will be responsible for implementation.
Peditto cautioned against allowing the process to become a series of demos without a clear understanding of who will ultimately approve the purchase.
Without clear ownership, HR teams can spend considerable time evaluating products without knowing whether the organization is actually prepared to buy.
Evaluate the Vendors
Once HR understands its needs and has established decision-making responsibilities, it can begin asking vendors more detailed questions.
Peditto emphasized that HR professionals don’t need to become AI engineers or understand every technical detail.
“You don’t have to solve the AI. You don’t have to build the AI,” he said. “But you need to know who does and where the information is coming from and who to call and who is responsible when issues arise.”
He also advised HR professionals not to be discouraged if sales representatives cannot immediately answer technical questions. The initial conversation should establish what needs further investigation. However, an inability to provide meaningful answers when HR reaches the vendor’s product or technical teams could be a warning sign, he said.
5. How is the tool built, and where does its data come from?
HR should understand whether a vendor developed its own AI models or built its product on top of an existing large language model, such as ChatGPT.
That distinction can have implications for reliability, security, and business continuity. If a vendor depends on another company’s underlying technology, HR should understand what happens if that provider experiences an outage or changes its services.
Training data is another critical consideration, particularly for tools used in hiring, compensation, and employee assessments.
“Where did the training data come from? This is a huge one,” Peditto said. “This is where bias liability is, where not knowing the answer could get you in trouble.”
The training data may also be irrelevant. For example, a compensation platform trained primarily on data from large corporations might produce recommendations that aren’t appropriate for a small business or startup.
Employers should also understand what happens to information they upload, including resumes, salary data, and assessment results, Peditto said. “What happens to that data? You will have to answer to an employee or a job seeker at some point. Is the data going somewhere else? Is it being kept private?” he asked.
6. How is the tool tested for bias, and can the vendor provide documentation?
Vendors should be able to explain how they evaluate their products for discriminatory outcomes and what documentation they can provide.
For hiring tools, HR should ask whether testing examines potential disparities involving protected characteristics and whether the vendor conducts ongoing evaluations as its models and underlying data change.
Employers also need to understand how vendors address applicable AI regulations. Requirements vary by jurisdiction, and organizations may face obligations involving independent audits, candidate notifications, impact assessments, and transparency.
Peditto said HR buyers should not expect to receive every answer during an initial demo. The important thing is to establish that these questions will be addressed before purchase.
He also warned that companies could face difficult questions in future litigation if they cannot explain how their AI tools were evaluated or what information informed their decisions.
7. Are companies similar to ours using this technology successfully?
Vendor claims about prominent customers can be impressive, but they don’t necessarily demonstrate that a product will work for every organization.
A platform used successfully by Amazon, for example, might not be appropriate for a small business with different hiring needs, resources, and workforce structures, Peditto said.
“You need to get references and reviews from companies like you,” he added.
HR shoppers should request references from organizations of comparable size and within similar industries. “Ask their customers what surprised them about the technology, what didn’t work as expected, and whether the vendor delivered on its promises,” he said.
Those conversations can provide a more realistic picture of implementation challenges, ongoing costs, and the level of support organizations can expect.
8. Can we see how the AI reaches its recommendations?
Transparency and explainability are essential when AI influences employment decisions.
If a platform ranks applicants, recommends employees for promotion, or evaluates employee performance, HR needs to understand the factors behind those recommendations.
Peditto described opaque systems as a form of black-box AI: Technology produces an answer, but users cannot see what happened behind the scenes.
If a hiring manager asks why one candidate was ranked above another, HR needs more than “because the AI told me,” as an answer.
HR should ask whether they can access the reasoning behind recommendations, override automated results, and adjust or retrain systems when necessary, Peditto said.
The ability to explain and challenge AI-generated decisions is important not only for internal accountability but also for responding to audits, employee complaints, and legal challenges.
Prepare for Implementation
The evaluation process doesn’t end when an organization selects a vendor. HR must also understand how the technology will be supported, monitored, and updated.
9. What happens when the technology goes wrong?
No AI system is infallible. Models can produce inaccurate recommendations, introduce bias, malfunction, or generate unexpected results.
HR buyers should ask vendors for examples of problems they have encountered, how those issues were resolved, and what escalation procedures are available, Peditto said.
He added that vendors may be reluctant to discuss failures, but their willingness to acknowledge problems and explain how they were addressed can be revealing.
“There is a difference between them saying that they will look into that, and saying that nothing has ever gone wrong,” he said.
He advised employers to establish their own internal response plans to potentially discriminatory recommendations. Contracts deserve particular attention, Peditto said. HR should work with legal and procurement teams to understand liability provisions, indemnification clauses, service commitments, and the vendor’s obligations when something goes wrong.
10. How will the technology evolve, and how much control will we have?
AI products are changing rapidly, and a tool’s capabilities, underlying models, and features may look very different even a year after purchase.
HR shoppers should understand how frequently vendors update their systems, how new data is incorporated, and whether updates undergo the same testing and validation as the original product, Peditto said.
“Is the latest data audited with the same care? What kind of notice do we get about new features? Do we get to test new features? Do we get to opt out of new features?” he asked.
“Flexibility can be hard for the vendor,” Peditto said. “But it’s worth having the conversation up front.”
He acknowledged that there are far more than 10 questions worth asking, but these provide a foundation for more informed purchasing decisions.
Ultimately, HR doesn’t need to become a technical department, Peditto said. But as AI becomes more deeply integrated into workforce management, HR professionals should become more confident technology buyers.
Was this resource helpful?
Validate your HR expertise
Earning your SHRM-CP credential makes you a recognized expert and leader in the HR field.
Related Content
Transform HR with mobile technology to streamline workflows, enhance communication, and boost employee engagement across distributed teams.
A new study found that talent development practitioners believe performance-based assessments are more effective at measuring employee learning outcomes than knowledge tests.
As AI makes deeper inroads into organizations, employee concerns about the technology continue to grow. Here's how HR leaders can address the fear of AI among their own staff, as well as in the wider organization.