Do you remember when people worked remotely due to COVID?

A couple of years ago, some companies announced that remote workers had to return to the office.

The reasoning behind this was that employees’ performance was better on-site than remote.

The data, though, doesn’t support that premise.

Companies from Gallup to Deloitte to McKinsey to the GAO to many others all show strong and better performance for remote workers compared to in the office.

On top of this, fewer distractions and more completed tasks appear with remote versus on-site.

Seeking lower attrition rates and happier employees?

Hubstaff found that 71% of employees had higher satisfaction compared to on-site employees.

Want another study?

Consider a study that found employees who worked remotely reported lower stress and better mental well-being. (79% and 82% respectively).

The adoption of companies that require employees to be in the office 100% of the time is increasing.

Take a look at any job site, and the numbers are significantly lower for someone to work remotely only versus hybrid or 100% on-site.

Even though that darn data again indicates a higher percentage of employees who would leave a company if they ever went to the office.

What about those cost savings?

Neat, found that on average. Employers will save around $11K (USD) per remote employee.

The environmental impact of on-site compared to remote operations has the potential to result in 54 million greenhouse gas emissions.

All of which, when viewed from a broader perspective, clearly indicate the benefits of remote versus on-site work, including environmental benefits, reduced employer costs, improved employee well-being, increased productivity, and fewer distractions.

Workforce efficiency and performance staring you, well, the person overseeing the whole strategy, right in their face.

Nevertheless, the push for full on-site improves productivity, and the benefits of such are overall the mode of operation for many companies, including Amazon.

I bring this up because the idea that AI will increase employee performance and make them more efficient isn’t, at this stage, a winning solution.

Let’s not kid ourselves here.

There will be plenty of companies, as data and studies show, that eliminate jobs and replace them with AI.

This isn’t in any stretch a benefit for employee performance.

Workforce efficiency, though?

McDonald’s experimented with AI to improve workforce efficiency, reducing the need for employees to take orders and serve food at the window.

It failed miserably.

An article in the New York Times discusses individuals with computer science degrees who focused solely on coding, only to lose their jobs or struggle to find employment due to coding being handled by AI.

One individual stated they applied to over a few thousand jobs to no avail. There wasn’t any mention of whether this person adapted, and while looking, started to take courses on AI programming – a necessity for the next generation of the workforce.

Schools and higher academia are finally recognizing the relevance of AI for tomorrow’s workforce and are moving away from the wholly coding experience.

However, for graduates with an AI programming degree, finding a job depends on many factors, which isn’t the focus of this thread.

Nor is the amount of money that will be thrown at them (assuming their talent level is elite).

What is missing in all these discussions is an effective business plan that incorporates AI into the development of workforce performance and efficiency.

A topic I will refrain from discussing in detail with this post.

The post here, instead, is about learning technology needs towards workforce efficiency and its rollout of functionality in learning systems (under the umbrella of LT), to meet those needs.

Are learning systems today meeting the needs of workforce efficiency?

The answer is no.

This is not unexpected, because of where AI stands, what is unknown (a lot), and what are its weaknesses and failures to solve (hallucinations, AI bias, and lately LLM traits moving over to another LLM), to say the least.

All of these exist, and yes, could be sitting in the system – not the bias – that is the client, ditto on the LLM we have, and we want to go into your system, and even the direct source, which may not exist – it’s your content, my dear client.

Systems themselves are limited in what they can do, even though, to their respective side, they are trying (as a whole) to induce and improve workforce performance and thus efficiency.

Is that learning’s role?

It is something we often fail to ponder.

What is the role of learning and training online in workforce performance and efficiency?

To guide? Yes

To offer opportunities to learn new skills and develop current ones, too? Yes

To empower an employee to produce more by knowing more?

All relevant only as it relates to what the company, i.e., the client, is offering them in terms of learning.

If we recognize that well-being is crucial for boosting productivity, offering employees the chance to take courses/content that are not only job-specific but also tailored to their personal needs makes perfect sense.

Yet, that is rare on the L&D and HR side. Training in its methodology finds the usefulness of personal and professional development.

That said, if the company forbades this mechanism of learning, then it is a negative on the part of the company itself.

There is by far nothing more stressful than the fear that AI will replace the employee.

That’s the stress a company should focus on from a learning standpoint. And make no mistake, there is real fear and thus stress here.
I stumbled upon a Reddit thread discussing the release of GPT-5 and its potential to replace white-collar workers.

The hype surrounding version 5 was driving this, as the respondents were not focusing on AI in general; some were, but overall, it was the latest LLM from OpenAI.

Even though there were mentions of the Ph.D. spin being incapable of solving basic math, there was also the occurrence of hallucinations and just general underwhelming sentiment.

Nevertheless, let’s say your company decided to roll out 5 in your workplace.

Under the presumption that the latest LLM would improve workforce efficiency and performance.

If the employee believes online learning will replace them and there is no plan to alleviate this fear, then the effectiveness of tasks becomes a serious question of what is defined as success.

There is, without a doubt, a time when two types of autonomous agents will appear within learning technology: free for XYS and, if you want more capabilities, a fee-based agent.

If I were a vendor investing in autonomous agents, I would consider this, depending on the agent’s level and what it can accomplish for the learner.

And that is a big piece, right?

For the learner.

Not just the admin, which would be of use to them.

Nor the manager or department head, again, depending on what it is doing and providing in terms of performance for the person in that role in the system.

But the learner.

Workflows and AI

It makes 100% sense, and the level to which it will be extensive or has the potential to be is a key element.

With an autonomous agent, you have one type of workflow, with learning and what “tasks” should be done or pushed out once X or Y is met is another, but what exactly is the task for the learner?

More importantly, should a task for an autonomous agent be handled by the learner, or should it be something the learner needs to learn themselves?

It’s one thing for AI to handle the creation of a slide deck (still requires a human to review, even if folks are ignoring doing this). Still, it is entirely different for someone who needs to learn how to do certain features, if not specifics, with PowerPoint, or even just the basics.

Would you want AI to handle this task in your system for that learner?

Answer engines will continue to improve, but their effectiveness is directly tied to the LLMs driving them, the multimodal approach (beyond just text, including audio, video, and PDF summaries) within the engine, the content, and the sourcing.

Today, Answer Engine for the most part is a search engine with some panache (as a whole).

It still requires the human element, as with any AI, until the days of Singularity happen, and impacts far more so than learning, but in our case here, learning.
Right now, though, how much should AI exist within a learning system, let alone another learning technology that improves the learner’s experience?

An answer engine surely would be – because knowledge management, at least in my opinion, will take an LMS, and other types of learning systems, to a whole new level.

If you look at it from a space level, what is a learning system?

It has always aimed to be, and continues to be, a platform for knowledge acquisition and growth, even if not explicitly stated.

With the engine, we can now say it out loud – the beginning of a new tier of knowledge management.

What then is the role of AI in a learning system from a performance standpoint?

Performance is the point here, because how we define performance with learning is the stuck in the mud, ROI thing, any L&D and Training exec will tell you, is data manipulation – after all, the necessity of saying 35 hours of learning reduced safety accidents by 84% or 42 hours of training (online) by ABC increased their sales by 43%.

Sure, doable, but there are always additional variables.

Nothing is as cause and effect, one-to-one.

With AI thrown into the learning mix, a whole new variable is at play.
Again, I am looking at the learner side and only their side.

Should I have AI handle tasks that learners typically perform or tasks that the AI can or will handle?

If it can read a publication and display only the highlighted text that correlates with the learner’s search, then yes, it helps in that way.

If the whole article is of zero interest to the learner, then I’d argue no.

Efficiency is often interwined with the guise of speed.

It is poor to think in such a way.

Especially with learning.

It reminds me of those “Speed Reading courses by Evelyn Wood,” that were prominent in the 70s and 80s.

OR those now thirty-somethings who would say that speeding through a YouTube video increased their learning comprehension and retention far more than the usual speed.

There is no mention of validation, say in six months, nor synthesis at 12 months.

Bottom Line

Visit any site these days that presents an article or post, including well-known publications, and you will see the summary above.

The summary includes specific points of relevance (as defined by the AI) for the reader.

However, the internet and its content are not primarily about learning to boost employee productivity in the workplace, whether they are on-site or remote.

Because, despite what you might think, on-site employee XTFD still could be reading on their mobile device, and not using their workplace desktop or laptop.

They are still not entirely focused on the task at hand.

Unless that task is about reading an article

Then yes, it is delivering.

But online learning on content that empowers the employee’s ongoing development of skills, and various personal and professional knowledge, shouldn’t be limited to an agent
.
Handling a task of this or that.

Instead, it should be in conjunction with the learner.

Symbiotic in a way

That maximizes

Rather than detracting.

E-Learning 24/7