How to Keep Up With AI: What People Do [2026]

How to keep up with AI is a question most people answer by doing, not reading. AISA research asked 84 people starting the AI fluency assessment "How do you typically learn about new AI capabilities?" and the top answer was hands-on experimentation: 39.3% said "I experiment myself." Social media came a close second at 35.7%, while colleagues, newsletters, and not following AI at all trailed behind. The results paint a clear picture of how people actually stay current, and they suggest a practical routine anyone can adopt.

1. Social media35.7%30
2. Colleagues13.1%11
3. Newsletters7.1%6
4. I experiment myself39.3%33
5. I don't actively follow it4.8%4

84 people answered, 25 August to 31 August 2026. Source: AISA research, aisa.to

How Do People Keep Up With AI?

Most people learn about new AI capabilities by trying things themselves or scrolling social media. Together, those two channels account for 75.0% of responses. Colleagues (13.1%) and newsletters (7.1%) play supporting roles, and only 4.8% said they don't actively follow AI at all.

That breakdown matters because it tells us something about the gap between formal training and self-directed learning. According to the Microsoft and LinkedIn Work Trend Index (May 2024), only 39% of people who use AI at work have had AI training from their company (source). BCG's AI at Work 2025 report, surveying over 10,600 workers in 11 countries, found that only 36% of employees feel adequately trained in AI use (source). When formal training is scarce, people fill the gap themselves.

The two dominant channels

Experimentation and social media serve different purposes. Social media surfaces what is new: a product launch, a viral demo, a capability people didn't know existed. Experimentation turns that awareness into understanding. The fact that both channels sit near 35-39% suggests many people likely use them in combination, even though each person picked one primary answer.

Where colleagues and newsletters fit

Colleagues at 13.1% and newsletters at 7.1% are lower-volume but often higher-trust channels. A colleague who says "I used this tool to cut our sprint planning time" carries more weight than a trending post. Newsletters, meanwhile, offer curation: someone else has filtered the noise for you. Both channels reward consistency over volume.

Is Social Media a Good Way to Keep Up With AI?

Social media is the second most common channel people use to learn about new AI capabilities, chosen by 35.7% of respondents. It is clearly useful for discovery, but it comes with trade-offs worth understanding.

What social media does well

Social media excels at speed and breadth. When a new model drops or a tool adds a feature, the first demonstrations usually appear on X, LinkedIn, or YouTube within hours. For someone trying to stay aware of what exists, that velocity is hard to beat.

Where social media falls short

The problem is signal-to-noise ratio. Feeds reward engagement, not accuracy. A flashy demo of a cherry-picked prompt gets more reach than a careful comparison of model benchmarks. If social media is your only channel, you risk confusing what is impressive in a demo with what is reliable in practice.

Making social media work for you

The practical move is to treat social media as a discovery layer, not a learning layer. When you see something interesting, open the tool and try it. That bridges the gap between the 35.7% who rely on social media and the 39.3% who learn by experimenting. You don't have to choose one channel; you can let social media tell you what to experiment with.

Why Do So Many People Learn by Experimenting?

Experimentation was the top answer at 39.3%, and it makes sense: AI tools are interactive by nature. You learn what a model can do by giving it a task and evaluating the output, not by reading a feature list.

Experimentation builds real understanding

Reading about context windows is one thing. Hitting the limit mid-conversation and watching the model lose track of your earlier instructions is another. Hands-on use builds intuition about what works, what fails, and where the boundaries are. That kind of knowledge is hard to get from a newsletter summary.

The risk of unstructured experimentation

The downside is that experimenting without a framework can leave blind spots. You might get comfortable with one tool and one type of task without realising there are entire categories of capability you haven't explored. Someone who only uses AI for writing may never discover its strength in task decomposition or data analysis.

A structured approach helps. Rather than experimenting randomly, pick a real task each week, try it with AI, and note what worked and what didn't. Over time, this builds a personal library of patterns that is far more useful than a bookmarked thread.

Experimentation and formal training are not substitutes

With only 36% of employees feeling adequately trained in AI use (BCG, June 2025, source), experimentation is filling a real gap. But it works best alongside some structure. Knowing what to experiment with, and how to evaluate the results, is itself a skill. Resources like the AI Fluency Academy's guide to prompting can give experimentation a backbone.

Do Newsletters and Colleagues Still Matter?

Yes, but they serve a different function than social media or experimentation. Newsletters (7.1%) and colleagues (13.1%) together account for 20.2% of responses, making them a meaningful minority channel.

Colleagues as a trust filter

When a colleague tells you about an AI capability, it usually comes with context: they used it for a specific task, in your shared domain, and can tell you whether it actually worked. That is a level of source triangulation you rarely get from a social media post. The 13.1% who rely on colleagues are getting curated, high-relevance information.

For teams, this has implications. If your organisation wants to accelerate AI adoption, making it easy for people to share what they've learned matters. A Slack channel, a monthly demo session, or even a shared doc of "things I tried this week" can turn individual experimentation into collective knowledge. Posts like AI Enablement: A Playbook dig into how to build that kind of sharing culture.

Newsletters as noise reduction

At 7.1%, newsletters are the least popular active channel. That doesn't mean they're ineffective; it may mean fewer people have found the right ones. A good AI newsletter does the filtering work for you: it reads the firehose so you don't have to. AISA publishes a weekly Latest AI News roundup that covers the developments worth knowing about, without the hype.

A practical comparison

Different channels serve different needs. Here's how they map to the learning cycle:

How Do You Stay Up to Date With AI Without the Overwhelm?

You build a routine that uses each channel for what it does best, and you set a time limit. The data shows people already gravitate toward one or two channels. The goal isn't to use all of them; it's to use the right ones deliberately.

A weekly routine that works

Here is a concrete routine built around the channels people actually use:

1. Monday: Read one curated source. A newsletter or the AISA Latest AI News roundup. Ten minutes. This tells you what changed last week.
2. Wednesday: Try one thing. Pick the most interesting item from Monday's read and use it on a real task. Thirty minutes. This is where the 39.3% who experiment are already spending their time.
3. Friday: Share one thing. Tell a colleague what you tried and whether it worked. Five minutes. This feeds the 13.1% who learn from colleagues, and it forces you to articulate what you learned.

Total time: under an hour a week. That is sustainable. The overwhelm comes from trying to follow everything in real time. A weekly cadence lets you stay current without the anxiety.

What to ignore

Not every AI announcement matters to you. A new model release is relevant if you use that model. A new tool is relevant if it solves a problem you have. Everything else is noise. Give yourself permission to skip it.

If you're unsure which skills to focus on, measuring where you stand helps. An AI fluency assessment gives you a baseline so you can direct your learning time toward the gaps that matter most, rather than chasing every new tool. Posts like How to Improve AI Skills: 7 Steps offer a structured path once you know your starting point.

Avoiding the "too many AI tools" trap

The feeling that there are too many AI tools is real. New products launch weekly. But most people only need two or three tools for their actual work. The rest is noise. Focus your experimentation on the tools you already use, and only evaluate a new one when you have a specific task it might solve better. The concept of tool selection criteria can help you make that call systematically.

What If You Don't Follow AI at All?

Only 4.8% of people said they don't actively follow AI. That's 4 out of 84 people. It's a small group, but it's worth addressing directly.

You're not as far behind as you think

If you haven't been following AI developments, the gap feels enormous. It isn't. The core capabilities of current AI tools, generating text, analysing data, summarising documents, writing code, have been stable for a while. What changes is the packaging and the edge cases. You can catch up on the fundamentals faster than you expect.

The Am I Behind on AI? post addresses this feeling directly with data on where people actually stand.

A starting point for the 4.8%

If you're in this group, here's a minimal starting point:

1. Pick one tool. ChatGPT, Claude, or Gemini. It doesn't matter which.
2. Give it a real task. Something you'd normally spend 20 minutes on: drafting an email, summarising a document, outlining a presentation.
3. Evaluate the output. Was it useful? What would you change? This is experimentation, and it's how 39.3% of people already learn.

You don't need to follow AI news to use AI effectively. But a small weekly habit, even just reading one curated roundup like AISA's Latest AI News, keeps you from being surprised when your organisation rolls out a new tool or policy.

The organisational angle

For managers reading this: if 4.8% of people aren't following AI at all, and only 39% have received company training (Microsoft and LinkedIn, May 2024, source), there's a gap between what organisations expect and what they provide. Measuring your team's current AI fluency before designing training ensures you meet people where they are, not where you assume they are. AI Training for Employees: Measure First makes the case for baseline measurement before spending on programmes.

About this research

This finding comes from AISA's own research. AISA asks a short research question of people as they start the AISA AI fluency assessment, one question at a time. The question in the chart above was asked from 25 August to 31 August 2026, worded exactly as shown and with the answers in the order shown, and 84 people answered it. Each person counts once: when someone answered more than once, their latest answer is the one counted.

More findings from this research are collected in AISA research and data.

How to cite this research: AISA (2026). "How to Keep Up With AI: What People Do [2026]". AISA research. https://aisa.to/blog/how-to-keep-up-with-ai

---

Related reading: Am I Behind on AI? [2026 Survey], data on whether people feel behind, and what the numbers actually show.

Related reading: How to Improve AI Skills: 7 Steps [2026], a structured path from baseline to proficiency.

Related reading: AI Training for Employees: Measure First [2026], why measurement comes before training spend.

Frequently Asked Questions

What is the best way to keep up with AI?

The best way to keep up with AI is to combine a curated weekly source with hands-on experimentation. AISA's research found that 39.3% of people learn by experimenting and 35.7% use social media. A sustainable routine uses a newsletter or roundup for discovery and dedicated time each week to try one new thing on a real task.

How do professionals learn about new AI tools?

AISA asked 84 people how they typically learn about new AI capabilities. The top channels were experimentation (39.3%), social media (35.7%), colleagues (13.1%), and newsletters (7.1%). Most people rely on self-directed methods rather than formal training, which aligns with BCG's finding that only 36% of employees feel adequately trained in AI use.

How much time should I spend keeping up with AI?

Under an hour a week is enough for most people. A practical split is ten minutes reading a curated roundup, thirty minutes experimenting with one tool or feature on a real task, and five minutes sharing what you learned with a colleague. The goal is consistency, not volume. Spending more time usually means you're consuming noise rather than building skill.