by admin | Aug 27, 2026 | Research & Productivity
A few months ago I caught myself doing something odd. I had used AI to draft a page for one of my sites, published it, and a week later I could not have explained the argument in it without opening the file. The work was done. The thinking, apparently, had not stuck.
That feeling has a name now, and researchers have started measuring it. The question of AI and critical thinking is no longer just a worry people raise at dinner. Three studies, two of them peer reviewed, have looked at what happens inside the heads of people who lean on AI, and the results are worth knowing before you build your next work habit around a chatbot.
The short version: the risk is real, it is not automatic, and the difference comes down to how you use the tool rather than whether you use it.
What the research says about AI and critical thinking
The most useful study of working adults came from Microsoft Research and Carnegie Mellon University, presented at CHI 2025. The team surveyed 319 knowledge workers and collected 936 first-hand examples of generative AI used in real work tasks.
The headline finding is a confidence effect, and it cuts both ways. People with higher confidence in the AI did less critical thinking. People with higher confidence in their own ability at the task did more. Same tool, opposite outcomes, depending on who was holding it.
The researchers also found that the thinking does not disappear so much as move. It shifts away from producing an answer and toward verifying information, stitching AI output together, and managing the task overall. That is still real cognitive work. It is just different work, and it only happens if you choose to do it. You can read the full paper on the Microsoft Research site.
The MIT essay study, and what it does not prove
The study that got the loudest headlines came from the MIT Media Lab. Researchers put EEG caps on participants and asked them to write essays under three conditions: using an LLM, using a search engine, or using nothing at all.
Brain connectivity was strongest in the no-tools group, moderate in the search group, and weakest in the LLM group. The LLM writers also reported the lowest sense of ownership over their essays, and struggled to accurately quote work they had produced minutes earlier. The authors called the pattern “cognitive debt”. The full paper is on arXiv.
Now the honest part, because this study has been badly oversold online. It involved 54 participants across the first three sessions and only 18 in the fourth, and it looked at one narrow task. It is still a preprint. The authors revised it in December 2025, but it has not been through peer review. That is a signal worth taking seriously, not a verdict on your brain.
The order you work in seems to matter
Buried in the same MIT study is the most practical finding of the lot, and almost nobody reported it.
In the final session the researchers swapped the groups around. People who had been writing unaided and were then given an LLM showed higher memory recall and stronger brain activation, closer to the search-engine group. People who had been leaning on the LLM and then had it taken away showed reduced connectivity and looked under-engaged.
Read that again, because it is the whole practical lesson. Thinking first and bringing AI in second looked healthy. Starting with AI and trying to think afterwards did not.
The 2026 study that names the difference
The newest piece of this puzzle is also the most useful, and it arrived in July 2026 in the peer-reviewed journal Frontiers in Psychology. Researchers surveyed 589 university students and early-career knowledge workers across three rounds spaced out over time, and they split AI use into two modes instead of treating it as one thing.
Dependent offloading is handing over the actual thinking: accepting the output with little examination, letting the AI structure your ideas, treating what it produces as the finished article. Autonomous offloading is using the same tool as a scaffold: taking the output as a starting point, comparing it against your own reasoning, and keeping ownership of the final result.
The two modes pulled in opposite directions. Dependent use was linked to handing over cognitive agency and to lower motivation, and through those to weaker self-rated judgement, creativity and depth of thinking. Autonomous use was linked to higher motivation and better self-rated outcomes. Same tools, same tasks, opposite results based only on how people engaged.
Then comes the finding that should change how you work. Both modes produced comparable immediate benefits. The session feels equally productive either way, which means unhelpful AI use is very hard to spot from experience alone. Nothing in the moment tells you which mode you are in. You can read the full study on the Frontiers in Psychology site.
Keep it in perspective. It asks people to rate their own thinking rather than testing it, so it shows associations, not proof of cause. But it is peer reviewed, far larger than the MIT study, and it names the distinction the other two keep circling.
Five habits that protect your thinking
- Write your rough version first. Even five messy bullet points before you open the chat window changes the whole session. You arrive with a position instead of asking to be given one.
- Ask the AI to question you rather than answer you. Try “argue against this” or “what am I missing here” instead of “write this for me”. You get pushback rather than a finished product you never examined.
- Verify anything you would be embarrassed to get wrong. Names, numbers, dates, citations, legal or medical claims. AI systems produce confident wrong answers regularly, which is why our guide to AI hallucinations is one of the most useful things to read before you trust an output.
- Explain the result out loud in your own words. If you cannot, you have not understood it, and you will not be able to defend it in a meeting or a viva.
- Notice when a task feels low-stakes. The Microsoft researchers found that people review AI output far less carefully when they judge a task to be unimportant. Low-stakes tasks are exactly where sloppy errors slip through into public work.
Important tip: if you cannot explain the answer to another person in your own words, you have not learned it. You have only borrowed it.
When letting AI do the work is completely fine
None of this means you should feel guilty every time you open a chatbot. Offloading is only a problem when you are offloading something you actually needed to learn.
Reformatting a messy list, fixing spelling, converting a table, writing boilerplate you have written a hundred times, summarising a document you are only skimming for one fact. Hand all of that over without a second thought. From my own experience running websites and online tools, that category is where AI saves genuine hours, and none of those hours were making me smarter.
The line is simple. If the task is the learning, do it yourself first. If the task is friction around the learning, automate it. That distinction is also why our post on using AI tools without cheating keeps coming back to the same test.
What this means for students and researchers
If you are studying, the stakes are higher, because the entire point of the work is to build something in your head that stays there. A summary you did not read leaves nothing behind. Our plan for using AI to study for exams is built around this: generate questions, not answers.
Microsoft Research has since argued for designing AI as a “tool for thought” rather than an assistant that hands you finished work, warning that people risk becoming validators of machine output rather than authors of their own. You can read their argument here. Until the tools are built that way, the burden of using them well sits with you.
Common Questions
Does using AI actually make you less intelligent?
No study has shown that. What the research shows is reduced mental effort during AI-assisted tasks and weaker recall of AI-assisted work. Those are measurable short-term effects on specific tasks, not evidence of permanent change.
Is it better to just avoid AI for serious work?
Avoiding it entirely costs you speed and puts you behind on a skill employers now ask for. The evidence points toward using it deliberately rather than avoiding it, which is also the theme of our guide to the AI skills that matter most for future jobs.
What is cognitive offloading?
It means using something outside your head to do mental work for you. A shopping list, a calculator and a satnav all count. AI is a far broader version of the same thing, which is why where you draw the line matters more.
Final takeaway
AI and critical thinking are not enemies. The research suggests they come apart only when you hand over the part of the task that was supposed to change you. Think first, use the tool second, verify what matters, and be able to explain the result without the screen in front of you. Do that and the tool stays a tool.
by admin | Aug 26, 2026 | Future Jobs
You spend an hour on an application. You tailor the CV, rewrite the cover letter, press apply. Then nothing, or a polite rejection that arrives so fast nobody could have read your work. The first question most people ask is the right one: did a human ever see this?
Some of it was almost certainly handled by software. That is what people mean by AI job application screening. This post covers what actually happens to your application, what the UK regulator found when it asked employers directly, the rights you already have, and the few things that genuinely help. I am not a lawyer, so treat this as plain English background rather than legal advice.
What AI job application screening actually looks like
Stop picturing one robot reading your CV and deciding your fate. The automation is spread across several small steps, most of them unglamorous.
- Form filters. Fixed answer questions: years of experience, right to work, notice period, location. A wrong answer here ends your application before a word of your CV is read.
- Parsing. Your CV is converted into structured fields. Columns, tables, headers and text inside images are where this goes wrong.
- Matching and ranking. Your wording is compared against the job description, and some platforms score candidates so the recruiter opens a sorted list rather than a pile.
- Assessments. Games, situational judgement tests and recorded video interviews, scored automatically.
You will see confident numbers online about how many CVs are thrown out by robots. Follow them back and they usually lead to a company that sells recruitment software. A figure with a better source behind it: the ICO, citing a survey by the Institute of Student Employers, notes that 70% of employers expect to increase their use of AI and automation in recruitment over the next five years. Growing fast, then, but not the sealed automated wall people picture.
The UK regulator asked employers what they were really doing
In March 2026 the Information Commissioner’s Office published Recruitment rewired, a report built on evidence from more than 30 employers who spoke to it voluntarily between March 2025 and January 2026. It names nobody, but the findings are blunt, and hardly anyone passed them on to jobseekers.
The headline finding is that many employers running automated recruitment are likely relying on solely automated decisions, meaning no meaningful human involvement, in decisions that significantly affect people. The ICO also found employers need to tell candidates much more clearly that automation is in use, and that where a human is involved, that involvement must reach everyone at that stage rather than a lucky few. It then wrote to 16 organisations it believed were making automated decisions about jobseekers, and all of them committed to acting on its recommendations.
Important tip: if you read only one link from this article, make it the ICO’s own page for jobseekers on automated recruitment decisions. It is short, written for candidates rather than lawyers, and it sets out exactly what you can ask for.
The rights you already have, in the EU and the UK
You do not have to wait for new AI laws. Article 22 of the GDPR has, since 2018, given people in the EU the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects or similarly significantly affects them. Where it applies, you can obtain human intervention, express your point of view, and contest the decision.
The UK version changed recently, and most job hunting articles still get this wrong. Section 80 of the Data (Use and Access) Act 2025 replaced Article 22 of the UK GDPR with new Articles 22A to 22D, in force since 5 February 2026. The old rule was a prohibition with exceptions; the new one is closer to a permission with safeguards attached, which is why the ICO now tells candidates plainly that employers may use automation if they have a valid reason. Article 22C keeps the substance: information about the decision, the ability to make representations, to obtain human intervention, and to contest it. That is the wording to use if you write to an employer about a rejection.
The two words that decide everything: solely, and meaningful
Both versions hang on the word solely. If a person genuinely took the decision, the rule does not bite, which is why employers keep someone in the loop. Article 22A says a decision is solely automated if there is no meaningful human involvement in the taking of it. A recruiter glancing at a ranked list and clicking reject is not obviously the same thing as a person deciding, and the ICO’s findings suggest that gap is common.
The ICO consulted on updated guidance until 29 May 2026 and the final version is due later this year, so for now the draft on its automated decision-making page is the clearest statement of what it expects.
What the EU AI Act adds, and when it actually starts
You may have read that tough new European rules on hiring AI landed in August 2026. That was the plan for a while, and many articles still say it. The European Commission’s AI Act page does classify AI tools for employment as high risk, naming CV sorting software as its example, and those systems will face obligations on risk assessment, data quality, logging, documentation, human oversight and accuracy.
The timing moved, though. Following the AI Omnibus, which entered into force on 27 July 2026, the Commission now states that the rules for high risk areas including employment apply from 2 December 2027. The AI Act itself became applicable on 2 August 2026, but the hiring obligations were pushed back, so the protection people are waiting for is still more than a year away.
What actually helps, in order
Rights matter for the unfair cases. Most of the time you simply want more replies. These are worth your energy, in rough order of payoff.
- Answer the form questions carefully. This decides more outcomes than anything else, and it is the part people rush.
- Keep the CV file boring. One column, real text rather than a picture of text, plain headings, consistent dates.
- Use the employer’s words. If the advert says management accounts and your CV says monthly reporting, add their phrase. That is not keyword stuffing, it is using the vocabulary the search uses.
- Do not fabricate to beat the filter. Hidden white text and invented job titles get you removed later and damage your name in a small industry.
- Find a human. One thoughtful message to someone on the team beats twenty more applications through the same form.
Using AI on your own side is fine, as long as you are the one thinking. Our guide to using AI in your job search covers the workflow, how to prepare for AI job interviews covers the recorded stage, and the AI skills that will matter most for future jobs are a better use of an afternoon than perfecting a CV template.
One caution from the security side of my own work. Application forms collect a lot: address, date of birth, salary history, sometimes identity documents. Check you are on the employer’s real careers domain rather than a lookalike, and never reuse an important password on a job portal. Our guide to using AI safely and protecting your privacy covers the rest.
Common Questions
Can I ask an employer whether AI screened my application?
Yes, and it is a reasonable question. In both the EU and the UK, where a significant decision about you was made solely by automated means, you are entitled to be told, to put your side, and to ask for a human to look again. Keep the message short and polite, name the role and the date, and ask whether a solely automated decision was involved and how to request human review.
Does a plain CV really beat a nicely designed one?
For anything submitted through an online system, yes. Two column layouts, text boxes and icons often parse badly, and a CV that turns into scrambled fields cannot rank well however good your experience is. Keep a designed version for people, a plain one for forms.
Is it acceptable to use AI to write my application?
Using it to structure your thinking, tighten wording and check you have answered the advert is fine. Do not let it invent experience, and never send anything you could not talk about confidently for five minutes in an interview. That one test catches most of the trouble.
Final takeaway
AI job application screening is real, but less mysterious and less all powerful than the headlines suggest. Much of what filters you out is a plain form question, not a clever model. You already have rights around fully automated decisions, the bigger European rules arrive in December 2027 rather than this year, and the UK regulator has told employers in writing that many are not doing this properly yet.
So do the boring things well, keep your CV readable by software and by people, ask a direct question when something feels wrong, and spend the time you save talking to actual humans. That last part is still the thing no system has automated away.
by admin | Aug 25, 2026 | AI Guides
You have a scan done, and then comes the quiet part. You wait. Somewhere in a hospital, a specialist has a long list of images to work through, and yours is one of them. That waiting is the reason so many hospitals are now testing software that looks at scans before a human does.
This post explains how AI reads medical scans in plain English. No hype, no scary robot doctor stories. Just what the technology actually does, where it is already being used, what it still cannot do, and what any of it means for you as a patient or a student.
What “reading a scan” means for a computer
A radiologist looking at an X-ray or an MRI sees shapes, shadows, textures and patterns, and compares them with thousands of images they have seen before. A computer cannot see any of that. To software, a scan is just a grid of numbers, where each number is the brightness of one tiny square of the image.
So the job of a medical imaging model is to find patterns in those numbers that line up with something a doctor cares about. A dense patch here. An unusual outline there. A change compared with the same person’s scan from two years ago.
How AI reads medical scans, step by step
Most medical imaging tools follow roughly the same path:
- Training on labelled scans. The model is shown a very large set of past scans where experts have already marked what was there and what was not.
- Learning the pattern. Over many passes, it adjusts itself until its answers match the expert labels more and more often.
- Testing on scans it has never seen. This is the honest test. A model that only performs well on its own training data is useless in a real clinic.
- Flagging, not deciding. In practice the output is usually a highlighted region, a score, or a priority ranking that pushes urgent cases up the queue.
- A human signs off. A qualified clinician still reviews the case and makes the actual call.
That last point is the one most news headlines skip. Almost none of these tools are approved to diagnose anyone on their own. They sort, measure, highlight and prioritise, and a person decides.
Where this is already happening
This is not a future prediction. It is already in regular use, and you can check that yourself rather than taking anyone’s word for it.
The US Food and Drug Administration publishes a public list of AI-enabled medical devices that are authorised for marketing in the United States. It is a long list, and if you scroll it you will notice how much of it sits in one category: radiology. Imaging is where this technology found its first real footing, because scans are digital, standardised and produced in enormous volumes.
In the UK, the NHS is running a large trial called EDITH, short for Early Detection using Information Technology in Health. According to the official government announcement, nearly 700,000 women across 30 sites are taking part, backed by 11 million pounds of funding through the National Institute for Health and Care Research. The question it is testing is very specific. Breast screening currently needs two specialists to read each mammogram. Can AI safely take the place of one of them, so the second specialist is freed up for other patients?
Notice the framing there. The goal is not a machine that replaces radiologists. It is a machine that absorbs some of the repetitive reading so that scarce human attention goes where it counts.
What AI still cannot do with a scan
Being honest about the limits is what separates a useful tool from a risky one.
- It does not know your story. Your symptoms, family history, medication and previous illnesses shape what a scan means. The model usually sees only the image.
- It can inherit bias from its training data. A model trained mostly on scans from one type of machine, one hospital or one population can perform worse elsewhere.
- It can be confidently wrong. This is the same failure mode you see in chatbots, described in our post on why AI gives wrong answers. A wrong answer does not arrive with a warning label attached.
- It struggles with the rare. Unusual conditions have few examples to learn from, which is exactly where a human expert’s judgement matters most.
Why the explanation matters as much as the answer
Here is the part that gets the least attention and deserves the most. If a model highlights a region of a brain scan, a doctor’s very next question is “why”. Without an answer, the clinician is being asked to trust a black box on something that affects a real person’s treatment.
That is the whole field of explainable AI, which tries to show which parts of an image pushed the model towards its conclusion. It is one of the most active research areas in medical AI right now, and it is the difference between a tool clinicians actually adopt and one that sits unused. We cover the basics in what is explainable AI.
Important tip: when you read any claim about medical AI, look for two things. Was it tested on patients it had never seen before, and does a qualified human still make the final decision? If a story cannot answer both, treat it as marketing rather than medicine.
From my own experience building websites and working around cybersecurity, the pattern is familiar. The tools that survive in serious settings are never the flashiest ones. They are the ones that log what they did, show their working, and fail in a way a human can catch. Healthcare is that principle turned up to maximum.
What this means for you
If you are a patient, the practical effect is mostly speed. Faster triage of urgent cases, shorter queues, and a second set of eyes that never gets tired at the end of a long shift. You are still being cared for by people.
If you are a student or early researcher, this is one of the most open fields going. It rewards people who understand both the technical side and the clinical reality, and the shortage is real. Our guide on using AI for research and productivity is a reasonable starting point if you are heading in that direction.
And if you are simply curious, the World Health Organization has published guidance on the ethics and governance of AI in health, with more than 40 recommendations for governments, technology companies and healthcare providers. It is a readable window into the questions the people running this stuff are actually arguing about.
Common Questions
Can AI diagnose a disease from a scan by itself?
In routine practice, no. Approved imaging tools generally assist a clinician by flagging, measuring or prioritising. A qualified person reviews the case and makes the diagnosis.
Is AI better than a radiologist?
That is the wrong comparison. On a narrow, well defined task with plenty of training data, a model can be fast and consistent. On the wider job of interpreting a scan in the context of a whole person, a trained specialist is doing something the model is not even attempting.
Are my scans used to train AI?
Rules differ by country and by hospital, and medical data is tightly regulated in most places. If it matters to you, ask your healthcare provider directly what their data policy is. That is a fair question and they should be able to answer it.
How do I start learning this field?
Begin with the fundamentals of machine learning rather than with medical AI specifically. Our beginner explainer on what AI actually is is a gentle first step, and the maths and coding basics come next.
Final takeaway
Understanding how AI reads medical scans mostly means letting go of the dramatic version of the story. There is no machine sitting in a dark room deciding who is ill. There is software that spots patterns in pixels, hands its best guess to a human, and gets checked. The interesting work now is not making it cleverer. It is making it explain itself well enough to be trusted, and testing it honestly enough to be safe.
This article is for general education only. It is not medical advice. For anything about your own health or your own scan results, speak to a qualified healthcare professional.
by admin | Aug 24, 2026 | AI Tools
Your browser used to be a window. You typed an address, a page appeared, and that was the whole relationship. In 2026 the browser wants to talk back. It offers to summarise the page you are reading, answer questions about it, and in some cases click buttons on your behalf while you go and make tea.
That shift has a name, and it is showing up in a lot of headlines without much plain explanation. So here is the simple version: what an AI browser actually is, which ones you can use right now, what they are genuinely good at, and the one security habit worth building before you hand any of them your logged-in accounts.
What is an AI browser?
An AI browser is a normal web browser with an assistant built into it. Instead of copying text into a separate chat window, the assistant already sees the page you have open, so you can ask about it directly.
Most of them do three levels of things:
- Reading and answering. Summarise a long article, pull the key numbers out of a report, explain a paragraph in simpler words.
- Working across tabs. Compare two products you have open, or gather details from several pages into one list.
- Acting for you. The assistant actually navigates and clicks. Booking something, filling a form, updating a repeat order. This is the part usually called agentic browsing, and it is the newest and least mature layer.
The first two are useful today and fairly low risk. The third is where the interesting promises and the real caution both live.
The AI browser options in 2026
The landscape moved quickly, and one of the biggest names has already left it.
ChatGPT: browsing moved back inside the app
OpenAI’s standalone browser, Atlas, stopped working on 9 August 2026. OpenAI’s own deprecation notice says browser-based agentic work is moving into ChatGPT and Codex instead, with the ChatGPT desktop app pointed to as the place for that work and a Chrome extension or sidebar for lighter help while you browse. We covered the announcement and what Atlas users needed to save in this post on the Atlas shutdown.
Gemini in Chrome
Google has been folding Gemini directly into Chrome through 2026, expanding country by country. As of Google’s 18 August 2026 announcement, it is available to all Android users in the US, and AI Pro and AI Ultra subscribers there also get “auto browse”, which handles tasks like booking parking or reorganising a repeat order. Availability outside the US still varies, so check before you assume the feature is missing on your device.
Perplexity Comet
Comet is a Chromium browser built around Perplexity’s answer engine, with an assistant sidebar that can act on pages. Perplexity made it free to download worldwide in October 2025 rather than keeping it behind its top subscription tier, which is why it turns up so often in “best AI browser” lists. It is available from perplexity.ai/comet.
Copilot Mode in Microsoft Edge
Edge has an opt-in Copilot Mode that adds a chat-first new tab, a Journeys view that groups your past browsing into projects, and Copilot Actions that can carry out tasks on sites. The action features have rolled out gradually and some are limited by region and plan, so what you see depends on where you are and what you pay for.
What AI browsers are actually good at
Strip away the launch videos and a fairly honest picture emerges. AI browsers earn their keep on reading-heavy work.
- Long documents. Terms of service, council planning notices, a forty page PDF you only need one section of.
- Comparison shopping and research. Six tabs open, one question: which of these actually has the feature I need?
- Second language browsing. Reading a site in a language you are still learning and asking for the meaning without leaving the page.
- Getting back into a project. Features like Journeys are genuinely handy if your research habit is “open thirty tabs and abandon them”.
Where they still disappoint is anything requiring judgement about accuracy. The assistant will summarise a wrong page just as confidently as a right one. It has no idea that the blog you are reading is three years out of date. That check is still yours.
From my own experience building sites and testing tools, the summaries are also quietly lossy in a way that matters for work. They tend to drop the caveats, the exceptions and the “unless” clauses, which are usually the sentences you actually needed. For anything with money or a deadline attached, read the source paragraph yourself.
The security habit worth building first
Here is the part that gets skipped. A normal browser reads a web page as content. An AI browser reads it as instructions too, and it cannot always tell the difference between what you asked and what the page told it to do. Attackers hide text on a page to hijack the assistant, which is why prompt injection is the defining security problem of this whole category.
The vendors know it. Google’s Chrome announcement explicitly says its models are trained to detect known threats including prompt injection, and that auto browse asks for confirmation before some sensitive tasks. That is reassuring and it is also an admission that the risk is real and not fully solved.
Important tip: never let an AI browser act while you are logged in to your bank, your email or your hosting control panel. Use a separate browser profile with no saved passwords and no payment cards for anything agentic, and keep your signed-in life in a plain browser.
Working around cybersecurity, this is the same instinct as not doing admin work from your daily account. It costs you thirty seconds to set up a second profile and it removes almost all of the worst-case outcomes.
How to try one without regretting it
- Start read-only. Spend a week using it only for summarising and questions. No actions, no logins.
- Check one summary a day against the source. You will quickly learn where it cuts corners.
- Watch the first few actions all the way through. Do not walk away from an agent on its first booking.
- Read what it stores. Browsing memory is a feature and also a record. Know how to clear it.
- Keep a plain browser. There is no rule that you need one browser for everything.
If you are still deciding which assistant suits you before worrying about which browser it lives in, our comparison of ChatGPT, Gemini and Claude is the better starting point.
Common Questions
Do I need to pay for an AI browser?
Not to start. Comet is free to download, Copilot Mode in Edge is free to switch on, and Gemini in Chrome is included where it has rolled out. The agentic features are the part most often tied to a paid plan.
Is an AI browser safe for online banking?
Treat it as not safe for that. Use a plain browser for banking and keep the AI browser for reading and research, especially if you have any agentic features enabled.
Will an AI browser replace search?
Not yet, and not entirely. It changes how you read results more than whether you search. You will still search, you will just spend less time opening ten tabs to find one sentence.
What happened to ChatGPT Atlas?
It was discontinued and stopped working on 9 August 2026. OpenAI moved browser-based agentic capability into ChatGPT and Codex rather than continuing a separate browser.
Which AI browser is best for students?
For coursework reading, any of them will summarise well, but a purpose-built research tool is often better. Compare with these AI tools for daily work and study before committing to a new browser.
Final takeaway
An AI browser is not a different internet. It is a normal browser with a fast reader sitting beside you, and that reader is genuinely useful for the boring parts of the web. Use it for summarising, comparing and understanding. Be slow and deliberate about letting it click things. Keep one plain browser for your signed-in life, and you get most of the benefit with very little of the risk.
by admin | Aug 23, 2026 | AI Guides
You open a coding tutorial, get stuck on line twelve, paste the error into an AI chat, and four seconds later you have working code. The program runs. You feel like you learned something. Two weeks later you cannot remember how any of it works.
If that sounds familiar, you are not lazy and you are not bad at this. You have just found the trap that sits at the centre of every attempt to learn to code with AI. The tool is genuinely good at producing answers, and answers are not the same thing as skill.
This guide is the practical middle path. You can use AI while you learn, and you probably should, but only if you change how you ask.
AI has quietly become a coding teacher
This is not a small trend. In the 2025 Stack Overflow Developer Survey, which collected more than 49,000 responses from 177 countries, 44 percent of people learning to code said they used AI tools to do it, up from 37 percent the year before. Across all developers, 84 percent said they use or plan to use AI tools in their work.
Here is the part that rarely makes the headlines. In the same survey, 46 percent said they do not trust the accuracy of AI output, up sharply from 31 percent the previous year. So the people using these tools most are also the people getting more careful about them, not less.
Why just asking the AI quietly backfires
Reading correct code feels like understanding it. Your brain recognises the shape, nods along, and files it away as learned. Then you open a blank file and nothing comes out. Recognition is not recall, and only recall gets you through an interview, an exam, or a broken project at eleven at night.
The Stack Overflow data backs this up from the other end. Forty five percent of developers said debugging AI generated code is time consuming, and when asked why they would still want to ask a human even if AI could write most code, 61 percent said they want to fully understand their own code. You cannot fix what you never understood.
From my own work building websites and running online projects, the pattern is consistent. Code I pasted in without reading is the code that broke six months later and took an entire evening to untangle. Code I wrote badly myself, then improved, I can still explain today.
The one rule that fixes most of it
Try for ten minutes before you ask. That is the whole rule.
Ten minutes of genuine struggle is where learning actually happens. You form a guess, you test it, you are wrong, and your brain marks that spot as important. If you ask the AI at minute one, you skip the part that does the work. If you are still stuck at minute eleven, asking is completely reasonable.
Important tip: set an actual timer. Ten minutes feels like an hour when you are stuck, and without a timer most people give up at ninety seconds.
Five ways to use AI that build skill instead of replacing it
- Ask it to explain, not to write. Instead of write me a function that sorts a list, try explain how sorting works in Python and show me one small example I can type out myself.
- Ask for a smaller exercise. Say I am stuck on loops, give me three tiny practice problems that get harder, and do not show me the answers yet. Then actually do them.
- Paste your own broken code first. Ask what is wrong with this and why, rather than fix this. The why is the lesson. The fix is just today.
- Make it quiz you. At the end of a study session, ask it to test you on what you covered, with questions that need you to write code rather than pick an option.
- Translate error messages. Error text is written for people who already know the language. Asking what does this error mean in plain English is one of the fastest legitimate uses of AI while learning.
Notice what all five have in common. The AI is doing the explaining and the coaching. You are still doing the typing and the thinking.
Keep a real course as your spine
AI is a brilliant tutor and a terrible curriculum. It answers whatever you ask, which means the gaps you do not know about stay gaps. A structured course fixes that, and the best ones are free.
- freeCodeCamp is a nonprofit with full certification tracks in web development, Python and data analysis, all free.
- Harvard CS50x is the university computer science introduction, free to audit, and it teaches you how computers think rather than just one language.
- Microsoft Learn has a beginner Python path with exercises you run in the browser.
- Kaggle Learn offers short practical courses if you are heading towards data and machine learning.
Pick one and finish it. Finishing a mediocre course beats starting four excellent ones. If you want a wider list with the AI specific options included, we compared them in our guide to free AI coding courses for beginners, and there is a broader roadmap in how to learn AI for free.
How to tell if you are actually learning
Use the blank file test. Once a week, close every tab, open an empty file, and rebuild something small you already did with help. A calculator, a to do list, a script that renames files. No AI, no tutorial, no copying.
If you can do it, the learning stuck. If you cannot, that is not failure, it is just useful information about which week to repeat. Most people who feel stuck are actually people who have never tested themselves without a safety net.
It also helps to know what you are aiming at. Coding, machine learning and AI are related but not the same job, and we broke down the difference in AI vs machine learning vs deep learning.
Common Questions
Can I learn to code using only AI?
You can get surprisingly far, but you will end up with holes you cannot see. AI answers the question you asked, not the question you did not know to ask. Pair it with one structured course and the holes mostly close.
Which language should I start with?
Python for almost everyone. It reads close to English, it is the main language of AI and data work, and its use jumped seven percentage points between 2024 and 2025 in the Stack Overflow survey. If you specifically want to build websites, start with HTML, CSS and JavaScript instead.
Is it cheating to use AI while learning to code?
Not on your own projects. On graded university work, check your institution rules first, because they vary and the penalties are real. The honest test is whether you could explain every line if someone asked.
How long does it take to get job ready?
Anyone promising a fixed number is guessing. Realistically, six to twelve months of consistent practice gets most people to a junior portfolio level. Consistency matters far more than daily hours. Forty minutes every day beats one heroic Sunday.
Final takeaway
The people who will struggle are not the ones who use AI. They are the ones who let it do the thinking. Use it as a tutor that never gets tired of your questions, keep one real course running underneath, try for ten minutes before you ask, and test yourself in a blank file every week.
Do that and you get the speed without the hollow middle. Open your editor today and write ten lines badly. That is a better start than reading one more guide.
