Independent guides to the best AI and language learning tools, compared on what they do, what they cost and who they suit.
Every week brings a new AI tool that promises to write your essays, build your app, design your logo or run your calendar. Some of these tools are genuinely transformative. Many are thin wrappers around the same handful of models. And almost all of them look better in a demo than they do on a Tuesday afternoon when you have a real deadline and messy material.
Top Tools is our attempt to cut through that. The Review NYU is an independent publication about AI tools and the modern AI world, and in this section we compare the main categories of tool, explain what they actually do, what the research says about where they help and where they do not, and how to test them yourself. We cover six areas: chatbots, writing, image and video, coding, productivity, and language learning, because learning a language is one of the most popular and most interesting uses of AI. We do not accept payment for placement.
| Section | What it covers | Start here if you want to |
|---|---|---|
| AI chatbots | General AI assistants and chatbot apps | Choose a main assistant and use it well |
| AI writing | Drafting, editing, rewriting and research writing tools | Write faster without losing your voice |
| AI image and video | Image generators, video generators and editors | Create visuals responsibly |
| AI coding | Code assistants, AI editors and agents | Ship code faster without shipping bugs |
| AI productivity | Meeting notes, email, research, automation | Save time on routine work |
| Language learning | AI tutors, apps, websites, books, films, songs, podcasts | Learn a language with AI as a study partner |
AI tools have moved from novelty to everyday use remarkably quickly. Pew Research Center's survey of 5,119 U.S. adults, conducted in February 2026, found that about half of U.S. adults now use AI chatbots, up from a third in 2024, and that about a quarter use them daily. The most common reasons were searching for information and, among employed adults, tasks at work. Pew also found that views about AI tilt negative, even among younger adults who use chatbots most. That combination, heavy use alongside real scepticism, is a good description of how we approach these tools.
The most useful single idea from AI productivity research is that AI's abilities are uneven. In a preregistered experiment with 758 Boston Consulting Group consultants, Fabrizio Dell'Acqua and colleagues found that on tasks inside the AI's capabilities, consultants using GPT-4 completed 12.2 percent more tasks, worked 25.1 percent faster and produced results rated 40 percent higher in quality. On a task designed to fall outside those capabilities, consultants using AI were 19 percentage points less likely to reach the correct answer. The authors call this the jagged technological frontier: tasks that look similarly hard can fall on very different sides of what AI does well.
Other studies point the same way. In a 2023 experiment published in Science, Shakked Noy and Whitney Zhang found that access to ChatGPT cut the time professionals took on writing tasks by 40 percent and raised rated quality by 18 percent. But in a 2025 randomised trial by METR, experienced open source developers working on their own large codebases took 19 percent longer when allowed to use AI tools, even though they expected to be faster.
| Study | Setting | Finding |
|---|---|---|
| Noy and Zhang, Science, 2023 | Professional writing tasks | 40% less time, 18% higher rated quality |
| Dell'Acqua and colleagues, 2023 | Consulting tasks inside AI's capabilities | 12.2% more tasks, 25.1% faster, 40% higher quality |
| Dell'Acqua and colleagues, 2023 | Consulting task outside AI's capabilities | 19 percentage points less likely to be correct |
| METR, 2025 | Experienced developers on their own repositories | 19% longer with AI tools allowed |
The lesson is not that AI is good or bad. It is that you need to find out, for your own tasks, which side of the frontier you are on.
Our reviews follow a simple principle: test on real tasks, not demos.
AI chatbots. Our AI chatbots section and the guide to chatbot apps explain how general assistants differ, what free and paid tiers include, how to write instructions that get good results and how to check answers.
AI writing. In AI writing and our guide to AI writing tools, we look at drafting, editing, rewriting and research writing, with particular attention to keeping your own voice and meeting academic and workplace rules.
AI image and video. The AI image and video section and the guide to image and video tools cover generators and editors, copyright, consent and labelling of synthetic media.
AI coding. In AI coding and our guide to AI coding tools, we compare autocomplete assistants, AI editors and coding agents, and explain why evidence on productivity is mixed.
AI productivity. The AI productivity section and the guide to productivity tools look at meeting assistants, email and document helpers, research tools and automation.
Language learning. Our language learning section covers apps, websites, books, films and series, songs and podcasts, with practical workflows for using AI as a study partner.
Most major AI tools offer a free tier and one or more paid plans. Free tiers are often enough to learn a tool and handle occasional tasks, but they usually limit how much you can use the most capable models, how long your files and conversations can be and which features you can access. Paid plans make sense when you hit those limits regularly on work that matters. Before paying, check whether your school or employer already provides access, because many institutions now license AI tools with stronger data protections than personal accounts.
Most people do not need a dozen AI subscriptions. A sensible stack looks like this:
| Layer | What it is | Why |
|---|---|---|
| One main assistant | A general chatbot you know well | Covers most tasks; skill with one tool beats shallow use of many |
| Specialist tools where they clearly win | Coding assistant, image generator, meeting notes | Only where they beat the main assistant on your tasks |
| Verification habits | Sources, tests, second checks | Catches the errors every tool makes |
| Data rules | What you will and will not paste in | Protects you, your employer and other people |
AI tools raise real questions about accuracy, privacy, copyright and fairness. The U.S. National Institute of Standards and Technology publishes an AI Risk Management Framework that organisations use to think about these risks. For individuals, a few habits cover most of it: do not paste confidential or personal information into tools unless you understand how it is stored and used, check facts and citations before relying on them, disclose AI use where your school, employer or publisher requires it, and respect other people's work and likeness when generating images, video or text.
The difference between disappointing and excellent results is usually the instructions. Give the tool context: who you are, what the task is for, who will read the result and what good looks like. Provide examples and source material rather than asking it to work from memory. Ask for a specific format and length. Break big tasks into steps and check each step. And when the result is wrong, say exactly what is wrong rather than starting again. These habits work across chatbots, writing tools, coding assistants and image generators.
AI tools change monthly, and it is tempting to try everything. A calmer approach works better: review your main tools every few months, read about major changes from reliable sources and only switch when a new tool clearly beats what you use on your own tasks. Skill with a familiar tool compounds over time; constant switching resets it.
These illustrative examples are not real people.
A graduate student uses one general assistant for brainstorming, summarising readings and checking drafts, a reference manager for citations and never pastes unpublished data into tools without permission.
A junior developer uses an AI coding assistant for boilerplate and tests, reviews every change and turns it off when working on unfamiliar parts of a large codebase.
A language learner practises conversation with an AI tutor daily and checks surprising grammar explanations with a textbook.
These are recurring themes from user discussions, not quotations from individuals.
The Review NYU is an independent publication. We base these guides on hands-on testing, companies' own documentation and published research. We label company claims as such, do not accept payment for placement and never invent statistics, reviews or credentials. Read about our editorial standards.
It depends on your tasks. Most people get the most value from one general assistant used well, plus specialist tools only where they clearly beat it.
On many tasks, yes. Studies have found large time savings on writing and consulting tasks, but AI can also slow people down or reduce accuracy on tasks outside its strengths.
Test it on your real tasks against your free options, count the total time including checking and keep it only if it clearly wins.
Only if you understand how the tool stores and uses data. Avoid confidential, personal or unpublished material unless you are permitted and protected.
On real tasks, against a baseline, looking for errors, counting total time and checking cost and data practices.
Language learning is one of the most popular uses of AI tutors and chatbots, and a good test of what AI does well and badly.
Every few months is enough for most people. Check what you actually use, what you pay and whether a new tool clearly beats your current one on your own tasks before switching.
Often yes, because institutional licences usually come with stronger data protections than personal accounts. Follow their rules on what you may use the tools for.
This guide is independent and editorially reviewed. Company claims are labelled as such, and we never invent statistics, reviews or credentials. Read about The Review, our editorial standards, or contact us.