AI at Work · The Human Signal · Future of Work Lab

AI is built to sound right, which is not the same as being right

Dan Klein, a Berkeley professor who builds AI systems, explains that today's AI is trained to sound convincing rather than to be correct. That is why it can hand you a confident answer that is quietly wrong, and why smooth writing is no longer proof that the facts under it hold. The skill that keeps its value is not producing the words, which the AI now does well, but being the person who checks whether they are true before they go out.

For youSet one rule this week, that nothing the AI writes for you leaves your hands until you or a second tool has checked the specific facts, names, and numbers in it.

Source: The AI Report ↗

AI at Work Skills & Reskilling

You can hand off the writing and still sound like yourself

Writing a post or a work update used to mean staring at a blank box until the words came. Now you can point an AI at a few things you have already written, answer a handful of its questions in your own rough words, and have it learn how you actually sound, then draft in that voice for you to trim. The writing gets handed off, but the voice and the choice of what to say stay yours.

For youGive Manus (manus.im, free tier) three things you have written and one messy paragraph of your real opinion, then judge its draft on a single question, does this sound like me, and keep editing until it does.

Source: The Rundown AI ↗

AI at Work Skills & Reskilling

Piling more instructions on your AI can quietly make it worse

When an AI slips up, the natural fix is to add another rule, then another. One writer who does this for a living found his stack of rules had grown so heavy that a task loaded over eighteen thousand words of instructions before it even started. When he tested a bloated set of good instructions against a short, clean brief, the short brief did the job every time and the padded one failed two runs out of three.

For youOpen the custom instructions or saved rules you have given your AI, cut them back to the few that matter, and re-test one real task, because a smarter AI does not make your old rules smarter.

Source: Nate's Newsletter ↗

AI at Work Skills & Reskilling

When you build with AI, the skill that pays off is naming exactly what you want

When you ask an AI to build or design something, the result is only as clear as the words you give it. A new free site collects the names of common design pieces so you can point to the exact thing you mean instead of waving at it. It turns 'that bar across the top that does the thing' into a precise term the AI can act on.

For youNext time you ask an AI to make something visual, open the free library at https://namethatui.com/ and grab the real name of the piece you want before you ask. The wider habit worth keeping is that as tools do more of the making, saying precisely what you want becomes the part of the job that stays yours.

Source: Name That UI ↗

AI Agents AI at Work

One person can now put a phone assistant on the front desk

Answering calls, screening who is worth your time, and passing the good ones through used to need a receptionist or a paid service. Now a single person can build a voice assistant with no code that picks up the phone, asks callers a few qualifying questions, and forwards the real leads to you. For a freelancer or a tiny business, that is a job you can hand off without hiring.

For youIf calls eat your day, try building a simple call-screening assistant this week using a no-code tool like xAI's Voice Agent Builder, but write the questions it asks yourself, because which callers count is exactly the judgment you do not want to hand over.

Source: xAI ↗

Demographics & Labor AI at Work

The bosses who said AI would cut jobs are quietly changing their minds

A year ago, many company leaders were saying AI would let them cut a lot of staff. That prediction is fading fast, with far fewer bosses now expecting big cuts. The fear was easy to announce and is turning out to be hard to deliver.

For youNext time you hear that AI is behind a round of job cuts, pause before you take it at face value. Companies often reach for AI as the tidy explanation when the real drivers are a slow market, past overhiring, or a business already in trouble, because blaming the technology sounds better than naming any of those. Treat AI as one possible cause among several, and look for what else is going on before you believe the headline.

In under 18 months, the share of bosses bracing for AI job cuts more than halved.

Data: Exponential View

Source: Exponential View ↗

Skills & Reskilling AI at Work

Now that anyone can make things, the rare part is having something worth making

Doing the work used to be the hard part: the skills, the money, the team, the permission. AI is making the doing cheap, so the bottleneck moves to something older, which is having something worth saying. One electrician with no coding background built a small tool that sells for 12.99 dollars and replaces a 500 dollar service call.

For youThink of one thing you have always wanted to make but assumed you lacked the skills for, and start it this weekend with AI handling the parts you cannot. The scarce skill now is not doing but taste, so practice choosing what is worth making, not just how to make it.

Source: a16z ↗

AI at Work Skills & Reskilling

Stop telling AI what to do, and start asking it what it would do

Most people use AI by giving it an order and taking back whatever it produces. Jack Dorsey, who runs the payments company Block, says he flipped this: he now asks the AI for a few suggestions first, then picks the best one himself. The small change keeps him making the decisions instead of rubber-stamping one answer.

For youOn your next task, ask the AI for three options with their trade-offs instead of one finished answer, then choose. Asking for a menu keeps the judgment in your hands, which is the part worth protecting as the AI gets better at the doing.

Source: Jack Dorsey ↗

Organizational Change AI at Work AI Agents

Your workplace runs on unwritten rules built for a world that is ending

Every team runs on habits nobody chose on purpose: the standing meeting, the sign off chain, the one person everyone checks with. Most of them exist because time, information, or trust used to be scarce, and AI is quietly erasing some of that scarcity. There is a new catch too: if a rule lives only in a senior colleague's head, an AI helper working beside you cannot follow it, so to your tools that rule may as well not exist.

For youTake the one team ritual you quietly resent, ask what shortage it was invented to fix, and check whether that shortage is even still real before you argue to keep or kill it.

Source: Nate's Newsletter ↗

AI Agents AI at Work

The AI stops being a website you visit and starts working inside your apps

You used to open a separate AI website, ask your question, then copy the answer back into your email, files, or slides. OpenAI's new ChatGPT Work now reaches across your files and desktop apps and does the task where the work already lives, and it is aimed at the hundreds of millions of everyday users, not just coders. As a taste of what that opens up, you can now describe a simple website in plain words and it builds the page, puts it online, and hands you a link to share.

For youGetting started is simple. Sign in to ChatGPT Work, point it at one file you are already working on, and ask it to do the next step for you instead of copying answers back from a separate chatbot.

Source: OpenAI ↗

AI at Work Skills & Reskilling

Let the AI build a rough draft first, then spend your judgment fixing it

The old habit is to plan a task fully before you start, but the faster move now is to let AI throw together a rough version in minutes and then react to what is wrong with it. Andrew Ng puts it plainly: the AI's output is cheap and your judgment is the gold, so spend your time deciding, not typing the first draft. Each time you correct it, write that decision into a running note the AI has to follow, so it does not lose the thread and make you say it twice.

For youOn your next document or plan, ask AI for a rough version before you feel ready, then keep a short decisions note it must obey, because your reactions to a real draft are sharper than any plan you write from a blank page.

Source: The Batch (DeepLearning.AI) ↗

AI at Work Organizational Change

Your boss thinks AI already changed your job. Your actual workday says otherwise.

A new survey from the document company Nitro found that 85% of senior executives say they have rolled out AI, but only about half of managers say it has reached their own daily work. The gap is less about the technology and more about who has the time, training, and permission to make it pay off on real tasks. That gap is also an opening, because the person who quietly gets AI doing a genuine piece of their job becomes the one who connects the two worlds.

For youPick one repeated task this week, make AI do the grunt of it, and write down the time before and after, because almost nobody above you has actually done that hands-on part yet.

Source: Nitro ↗

AI at Work AI and Society

AI is quietly becoming a confidant, and that is the part to handle with care

A new study from the Oxford Internet Institute finds that asking AI for personal advice has become normal, from relationships to health to a late-night vent. That makes it a private advisor, not just a work tool, even though it can sound completely sure of itself while being wrong. What stays yours is the judgment about who and what to trust with the things that actually matter.

For youNext time you are tempted to ask AI something personal, ask it for the questions you should be weighing rather than the answer, then take the real decision to a person who knows you.

Source: Oxford Internet Institute ↗

Evals & Benchmarks AI at Work

The scoreboard for AI coding turned out to be partly broken

Much of the excitement about AI getting better at work rests on test scores, like the leaderboards that rank AI at writing software. OpenAI checked one of the most cited coding tests and found roughly a third of its tasks were flawed, sometimes marking correct work as wrong, and pulled its own recommendation to use it. The lesson travels well beyond code: a confident number is not the same as the AI actually doing your job well.

For youWhen a tool claims it beats humans at something, judge it on one real task from your own week instead of the headline score, because on your work you are the benchmark that counts.

Source: OpenAI ↗

Organizational Change Skills & Reskilling AI at Work

Workers are ready for AI. Their bosses are not.

A new McKinsey piece names a gap many people already feel: employees are quietly using AI every day, but leaders admit they are not set up for a world where AI does real chunks of the work. Closing that gap, moving a team from scattered dabbling to deliberate impact, is a job waiting for someone. If you can see both the work and the tools, you can be that person.

For youFind one place where your team uses AI in a scattered, everyone-for-themselves way, and write a one-page note on how it would work if it were deliberate; that memo will do more for your standing than another certificate.

Source: McKinsey ↗

AI at Work Frontier Models

Talking to AI stops feeling like a walkie-talkie

Until now, talking to the AI meant taking turns: you spoke, it waited, then it answered. A new voice version listens and talks at the same time, so it feels like a real phone call instead of leaving voicemails, and the basic version is free. You can think out loud, cut in mid-sentence, or practise a language while it corrects you as you go.

For youOpen the voice mode in the ChatGPT app and run one real task by talking, like rehearsing a hard conversation you have this week, and notice what you catch by hearing the answer that you would miss reading it.

Source: OpenAI ↗

AI at Work Skills & Reskilling

The mistake is using AI for chores instead of hard problems

Most people point AI at small stuff, tidying notes or drafting a quick email, and stop there. A widely shared essay argues the real payoff is the opposite: aim it at the hard, judgment-heavy problem you have been avoiding and ask it what you are missing. The chores save you minutes, but the hard problems are what actually change your week.

For youGive the AI the messiest, most important question on your plate right now, ask it for three things you have not considered, then decide for yourself which one is worth your afternoon.

Source: Christine Zhu ↗

AI Agents AI at Work

The fix for AI mistakes is not better AI. It is a second set of eyes.

Waiting for AI you can fully trust is the wrong bet. Accountants caught dishonest clerks for centuries by checking every figure against a second one, and pilots cut errors with a simple checklist, so do the same with AI: let it do the task, then add one step that checks the result before anyone acts on it. One writer ran a crew of AI helpers for about eight dollars to build a website, and when one of them faked a dozen quotes, the checking step caught it, not him.

For youMake it a rule this week. Nothing an AI writes for you goes out until a second check, yours or another tool's, has seen it.

Source: Nate's Newsletter ↗

AI at Work Skills & Reskilling

You can now build a good-looking website by borrowing a designer's rulebook

Building a decent-looking site used to need design skills or money for a designer. Now you can grab a ready-made design guide from a free library, hand it to an AI helper on your computer, and have it build pages that follow that professional style. For a student or job-seeker, that means a real portfolio site this weekend instead of someday.

For youBrowse the free Refero Design library, pick a style you like, and have an AI build a simple personal site that follows it. https://styles.refero.design/

Source: Refero Design ↗

AI at Work Organizational Change

As AI records every meeting, protect what only humans do there

Soon every meeting will be recorded and written up for you, so taking notes is no longer the point. What is left for people is the part a transcript cannot do: setting a clear agenda, reading the room, drawing out the quiet voices, and naming the decision before everyone leaves. One founder who studied top companies runs fewer and shorter meetings, and guards a no-meeting day so people can actually do the work.

For youPick one recurring meeting this week, add a three-line agenda and an end-early rule, and let an AI note-taker handle the write-up so you can focus on the discussion.

Source: Charter ↗