Top tools

Independent guides to the best AI and language learning tools, compared on what they do, what they cost and who they suit.

Updated 2026-11-22 by the The Review NYU editorial team

AI tools without the hype

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.

How Top Tools is organised

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

How many people actually use AI tools?

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.

What the research says about where AI helps

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.

How we evaluate tools

Our reviews follow a simple principle: test on real tasks, not demos.

  1. Real tasks. We use tools on the kind of work people actually do, with messy inputs and real constraints.
  2. Baseline comparison. We compare against doing the task without AI, and against a general assistant, because a specialist tool should beat both.
  3. Error hunting. We look for confident mistakes, invented facts and subtle errors, not just impressive output.
  4. Total time. We count the time spent prompting, checking and fixing, not only generation time.
  5. Cost and limits. We check free tiers, usage caps and what happens to your data.
  6. Company claims labelled. When we rely on what a company says about its product, we say so.

Inside each section

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.

Free versus paid AI tools

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.

A sensible AI tool stack

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

Using AI responsibly

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.

Choosing tools in five steps

  1. List the tasks that take you the most time or that you most want help with.
  2. Try your main assistant first on each task and note where it falls short.
  3. Trial one specialist tool for each gap, using free tiers.
  4. Measure honestly: total time including checking, and quality of the result.
  5. Keep what clearly helps and cancel the rest.

Getting better results from any AI tool

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.

Keeping up without chasing every launch

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.

Example users

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.

Common mistakes

  • Judging tools by demos instead of real tasks.
  • Paying for several tools that do the same thing.
  • Trusting confident output without checking it.
  • Pasting confidential information into tools.
  • Assuming AI helps on every task because it helped on one.
  • Ignoring your school's or employer's AI policies.

What users commonly say

These are recurring themes from user discussions, not quotations from individuals.

  • One assistant I know well beats five I barely use.
  • AI saved me hours on first drafts but invented a citation once.
  • The coding assistant was brilliant on boilerplate and useless on our legacy code.
  • Free tiers were enough until I hit the usage caps.
  • Learning to check AI output was the real skill.

How we write these guides

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.

In this section

  • AI chatbots: The best AI chatbots compared: apps, with what each one does, what it costs and who it suits.
  • AI writing tools: The best AI writing tools compared: tools, with what each one does, what it costs and who it suits.
  • AI image and video tools: The best AI image and video tools compared: tools, with what each one does, what it costs and who it suits.
  • AI coding tools: The best AI coding tools compared: tools, with what each one does, what it costs and who it suits.
  • AI productivity tools: The best AI productivity tools compared: tools, with what each one does, what it costs and who it suits.
  • Language learning tools: The best language learning tools compared: apps, websites, books, movies and series, songs and podcasts that help you make real progress.

Frequently asked questions

What are the best AI tools right now?

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.

Do AI tools really make people more productive?

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.

How do I know if an AI tool is worth paying for?

Test it on your real tasks against your free options, count the total time including checking and keep it only if it clearly wins.

Is it safe to put my work into AI tools?

Only if you understand how the tool stores and uses data. Avoid confidential, personal or unpublished material unless you are permitted and protected.

How do you test AI tools?

On real tasks, against a baseline, looking for errors, counting total time and checking cost and data practices.

Why include language learning in an AI tools guide?

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.

How often should I review the AI tools I use?

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.

Should I use the AI tools my school or employer provides?

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.

Sources

  1. Americans and AI 2026: Chatbots, Smart Devices and Views on Impact, Pew Research Center, June 2026
  2. Navigating the Jagged Technological Frontier, Dell'Acqua and colleagues, Harvard Business School Working Paper 24-013, September 2023
  3. Experimental evidence on the productivity effects of generative artificial intelligence, Shakked Noy and Whitney Zhang, Science, July 2023
  4. Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity, METR, July 2025
  5. AI Risk Management Framework, National Institute of Standards and Technology, accessed November 2026

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.