For L&D teams already stretched thin, the pressure to adopt the latest AI tools without a clear strategy is real. But speed without direction leads to low learner engagement and a whole lot of shiny platforms collecting dust. We put our biggest questions to Anand Timothy, Co-Founder and Director of Learning Solutions at Abara LMS and Abara Author, to cut through the noise. His take is refreshingly grounded: AI can solve real problems in L&D, but only if you know what problem you’re actually trying to solve.
Where AI Is Creating Real Value and Where It Is Not
AI is everywhere in L&D conversations right now. From your perspective, where is it creating real value today, and where is it still more hype than reality?
Anand: I see three primary areas where L&D is leveraging AI. First, is the speed of content creation. Right now, it takes weeks or even months to get meaningful training content out. Scheduling with subject matter experts, putting together course outlines, getting them approved by stakeholders, and then actually writing content and getting them into courses. It just takes too long. The second is speed of translation and the use of AI audio. Finally, I see AI adding value by improving existing LMS platforms to deliver more meaningful content to learners. The existing LMS platforms are too cluttered and often hard to navigate. Either there is not enough meaningful content, or there is too much. L&D teams have spent their full budget on shiny LMS systems and thousands of courses, but usage is low. Learners need curated content that is meaningful to their needs and delivered on time, not weeks or months later.
If they can’t find value in their LMS, learners will seek their own answers. Learners are used to finding their answers with a few questions in ChatGPT, that’s the user experience L&D teams are competing with.
“Learners are used to finding their answers with a few questions in ChatGPT, that’s the user experience L&D teams are competing with.”
Where I see more hype than reality is agentic AI for digital learning. Vendors are pitching magic bullets, the promise of full automation, start an agent and your learners are trained. That’s not how it works. You cannot entirely remove the need for sound instructional design oversight of the process or human review of content from a subject matter perspective.
The Mistakes Organizations Make When Adopting AI
What are the biggest mistakes organizations make when they try to introduce AI into their learning & development activities?
Anand: I think L&D teams are under pressure from the business and learners to demonstrate that they’re adapting to technology disruption caused by AI. It’s panic, and many are getting swayed by the ‘next big thing’ in technology.
I’ve seen companies with 5000+ employees and one or two L&D team members trying to get it all done. One of the mistakes I see is that they end up choosing some technology today that promises magic without major human oversight.
Instead, L&D teams need to find solutions to today and tomorrow’s problems. They need to navigate the changing landscape, experiment, pilot, and fine-tune programs by keeping the human in the loop. Just taking on the next big complex platform that promises to do everything often ends up creating more complexity.
What a Good AI-Enabled Learning Experience Actually Looks Like
From the learner’s point of view, what does a good AI-enabled learning experience actually look like?
Anand: It’s not just L&D departments that are overwhelmed and short-staffed. Learners and managers are already stretched thin, often learning new AI tools on top of their jobs. More learning content is not the solution.
What learners need is meaningful and relevant content to be surfaced based on the challenges they face, delivered directly to them. The ability to find more meaningful content quickly for the leadership challenge they are facing, rather than getting overwhelmed with 5 different leadership courses. Or better still, getting the answer to the question they have in no uncertain terms. Not “the answer to your question is in these 5 documents”.
Learners need access to AI along with training on how to use it in an approved way in line with the company policies. If not, learners will come to rely on their private AI tools to find the answers. Most users are already running multiple LLMs on their phones. That is the reality.
Personalization at Scale Without Over-Automating
AI is often positioned as a way to personalize learning at scale. What does that look like in practice, and how does one avoid over-automation?
Anand: You avoid over-automation by providing access to up-to-date content and keeping a human in the loop. If the content repositories that you add to AI systems are curated and produce responses that clearly include source attribution, you are relatively safe. Having a good ‘human-in-the-loop’ review mechanism for your source content and allowing learners to access content quickly from such trusted sources is the way to go.
As far as personalization is concerned, AI must be role aware. Aware of skill and knowledge gaps, and also be aware of context. What are the tasks the person has been given? What are the challenges they have been asking questions about? Personalization means giving them the right set of options, not directing them to more locations where they could run additional searches.
Making Development More Efficient Without Losing Effectiveness
How can organizations use AI to make the development of learning more efficient without losing the effectiveness of the learning experience?
Anand: I think more efficient means producing faster, not overproducing. More is not always good. If you produce tons of AI-generated content, it will eventually be overwhelming. Improved speed in filling training gaps is what every corporate organization needs. So plugging real gaps faster and more effectively is the answer. Reduce the load on your content/subject matter experts and accelerate content development with AI assistance.
Learning content must still meet some fundamental instructional design checkpoints. A course with well-designed learning objectives, with content that is valid, engaging, interactive, and an assessment that maps to those objectives, along with being available just in time when the learner needs it, all culminates into a ‘good course’. Just improving these metrics is also a win. Those are evergreen principles.
Using AI to Improve Measurement and Real-World Impact
Measurement is still a challenge for many L&D teams. How can AI improve the way organizations assess learning effectiveness and real-world impact?
Anand: Successful measurement of learning isn’t just pass/fail stats or scores. It’s a measurement against meaning and effectiveness, against target levels of skills, KPIs (Key Performance Indicators) or KRAs (Key Result Areas).
Learning needs to address a problem or an opportunity to demonstrate real-world impact. The problem: if the knowledge or skills are below what they should be, then training to upskill the person is the solution. The opportunity: the person’s knowledge and skills are strong for their current role, but they need to improve their knowledge and skill to deliver on future roles, that training can improve the organization’s talent pipeline. The ability to measure both knowledge, skills, and abilities for today’s roles and future roles is just as important. Success is leveraging AI to do this measurement, fill gaps in knowledge, skills, and abilities quickly, with minimal lag.
AI in Compliance-Heavy Environments
In compliance-heavy or regulated environments, where can AI genuinely strengthen training, and where does it introduce new risks?
Anand: Even in highly regulated environments, these regulations are published, and the organization’s legal departments have access to these regulatory documents. There are also public sites where such regulations are readily available. There are no secret government or industry regulations. You can have AI review this material and help improve your training material. The only thing is you need the “human-in-the-loop” to validate it.
Rather than depend on legal teams to act as the only source of regulatory and compliance training, content authors can flip the order and do their own homework, organize content, and send it to legal for review and validation. I think legal teams would be happier too. AI has been known to hallucinate legal precedents, case law, and actual laws, but that’s because someone skipped the human validation process. No important piece of AI-generated content should go live to a customer or an employee without human oversight.
Turning Legacy Content Into Something Learners Will Actually Use
Many organizations are sitting on years of content and internal knowledge. How can AI help turn that into something more usable and accessible for learners?
Anand: I think the first step is to agree with legal on what can and cannot be shared with AI. Some things must never be shared with an AI, but other pieces of information can be shared with the correct technical guardrails to ensure it is never used by LLMs.
The second step is to have that tested (internally or with external help) and validated. Once you have both, you can connect your content repositories to leverage AI. You can connect entire repositories or curate and selectively feed AI the content (in batches) it needs to process (for the output you are seeking).
But once you have your data in there, you have to figure out how to train users to reach and access this content. Your systems must be able to clearly tell you if the content is being accessed from internal documents or pulling something off the LLM or the web. Clearly marked sources are important for such capabilities to ensure trust.
Next, how do you leverage this AI-enabled content to be repurposed for various uses – learning content, just-in-time performance support, a quick one-pager, and also perhaps short videos too. If the users want full sets of documents analyzed and want to ask a series of questions to improve their knowledge or clarify questions they have, it should be able to do that too.
Adopting AI Without Losing Quality, Trust, or Engagement
If a client said, “We want to adopt AI in L&D, but we don’t want to lose quality, trust, or engagement,” what would your advice be?
Anand: My advice would be in three parts. First understand ‘why’ you want to adopt AI. Is it for speed? Scale? To replace gaps in your current LMS? What’s the business problem you are trying to address? This identification of the need helps you stay focussed and stops you from jumping onto a viral bandwagon. Match the problem to the solution for a more robust and relevant outcome.
The next bit is to find a partner or a couple of trusted partners who can help you pilot and experiment and do not just offer you a tool but also consultancy services to help support the adoption of AI. You will need more than one platform and small L&D teams will definitely need some services to help you run the program as well. Just a tool is not enough.
Final Thoughts
What stands out across every answer Anand gives is the same throughline: the human still matters. AI can accelerate content creation and take pressure off overloaded teams but it cannot replace sound instructional design, meaningful curation, or the accountability of a real person signing off on what learners receive. The organizations getting this right are not the ones chasing every new platform. They are the ones starting small, solving a specific problem and scaling only what actually works. If your L&D team is feeling the pressure to “do something with AI,” let that be your starting point. Not the tool, but the problem it needs to solve.
Meet Anand Timothy
Anand Timothy is the Co-Founder and Director of Learning Solutions at Abara LMS and Abara Author. With decades of experience in corporate learning, he helps organizations build smarter, more efficient L&D programs without sacrificing the human elements that make training stick.
Abara LMS is a modern learning management system built for organizations that need clarity, simplicity, and results. Abara Author is an AI-powered eLearning course authoring tool designed to help L&D teams produce high-quality content faster without losing instructional integrity.
