Using AI at Work

The fear is that the model will absorb your work and hand it to a competitor. In almost every real case that is not what happens and not what will hurt you. Four things do. You broke a promise your employer made in a client contract you have never seen. The information left your organisation's control and now sits in a third party's systems under their retention and their jurisdiction. You published something wrong with your name on it. Or you used your personal account because it was already signed in, and the protections your employer paid for did not apply to you. That last one deserves a sentence of its own, because the companies publish the difference and almost nobody at work knows it. OpenAI's policy says individual users' content may be used to train its models, and separately that business users' inputs and outputs are not trained on by default. Same company. Same model. Opposite defaults, decided entirely by which tab you happened to be in. This book is the working set of habits that follows from that. The four categories that never go in, and the one question that catches all four. Why removing the names does not make a spreadsheet anonymous. How to get real work out of these tools by describing the shape of a problem rather than pasting its contents. The AI notetaker sitting in your meetings, capturing people who never agreed to it. The four kinds of output that fail most often, which are numbers, names, quotes and links. When to disclose that you used it. And what to do if you have already pasted something you should not have, which no other book bothers to tell you. Not a prompting guide, and not an argument against using any of it. A way to be certain about four things and relaxed about the rest.

• Why the training fear is the wrong fear, and the four risks that actually cost people
• Why the consumer and business tiers of the same product can have opposite data defaults
• The four categories that never go in, and the one question that catches all of them
• Why deleting the names does not make a spreadsheet anonymous
• Getting good output by describing the shape of a problem instead of pasting its contents
• The two separate problems with code, going in and coming out
• AI notetakers in meetings, and the consent question nobody asks out loud
• The four kinds of output that fail most: numbers, names, quotes and links
• When to say you used it, and when nobody expects you to
• What to actually do if you have already pasted something you should not have

• Anyone using AI tools inside a job, especially without a clear company policy
• People handling client material, personal data or anything under contract
• Managers being asked what the rules should be
• Anyone running AI notetakers in meetings
• People who have already pasted something and are quietly worried about it

1. 01Introduction: Frightened of the wrong thing
2. 02Part One: What actually happens to what you type
3. 03Part Two: What goes in, and what never does
4. 04Part Three: The output has your name on it
5. 05Part Four: The part that is about you
6. 06Closing: Six things worth keeping

لا توجد تقييمات بعد.