Newshot | A lawyer used AI to write a report, but billed full time. What should I do? | ||
|---|---|---|
| 3 points by isolli 49 minutes ago │ hide │ past │ favorite │ 4 comments | ||
| This is not a request for legal advice, and therefore the country matters little here. I write because I am interested in hearing opinions on the moral implications of something that is bound to happen more and more often. Context: A lawyer wrote a report for me. Not only was it was of poor quality, but he billed a surprising number of hours for it. I then noticed that one of the references was completely made up, typical of AI hallucination. I started pulling on the thread and slowly became convinced that he had almost certainly heavily relied on AI and inflated his hours. We went to the first judicial level, and the judge rejected out of hand the various strands of evidence that I presented. I feel like the deck is stacked against me (the lawyer also holds several prestigious positions), so I don't think I'll appeal. The risk is too high. However, Pangram since burst onto the scene, and essentially confirmed my suspicions (it flagged 1/3rd of the report as AI-generated, including the hallucinated reference). I know that the false-positive rate is higher for legal text, but I'm nonetheless more convinced than ever that I am correct in my assessment. What would you do in this situation? I'm annoyed that he got away with it, and that I don't have the time or the finances required to prove it. |
Context: A lawyer wrote a report for me. Not only was it was of poor quality, but he billed a surprising number of hours for it. I then noticed that one of the references was completely made up, typical of AI hallucination. I started pulling on the thread and slowly became convinced that he had almost certainly heavily relied on AI and inflated his hours.
We went to the first judicial level, and the judge rejected out of hand the various strands of evidence that I presented. I feel like the deck is stacked against me (the lawyer also holds several prestigious positions), so I don't think I'll appeal. The risk is too high.
However, Pangram since burst onto the scene, and essentially confirmed my suspicions (it flagged 1/3rd of the report as AI-generated, including the hallucinated reference). I know that the false-positive rate is higher for legal text, but I'm nonetheless more convinced than ever that I am correct in my assessment.
What would you do in this situation? I'm annoyed that he got away with it, and that I don't have the time or the finances required to prove it.
| silverFork 17 minutes ago │ next [–] There are no moral implications. Assume that all of them will use it and by the way they have used Ai for a decade before this... way before it was even a mature discipline. In some countries its mandatory now. Germany I believe uses some IBM Ai system and few others use Ai where the case loads were piling up. They used call it expert system even years before Ai and it was designed to reduce their work load and improve the consistency and now its more sophisticated but still error prone. One should set the fees based on how fast and accurately they can deliver I suppose. If they made an error that costed you can make an effort to deal with them. There is no excuse for poor quality and low effort even though they have access to Ai. If there is a place to review their work, leave a review and specify the specific type of error. reply | silverFork 17 minutes ago │ next [–] There are no moral implications. Assume that all of them will use it and by the way they have used Ai for a decade before this... way before it was even a mature discipline. In some countries its mandatory now. Germany I believe uses some IBM Ai system and few others use Ai where the case loads were piling up. They used call it expert system even years before Ai and it was designed to reduce their work load and improve the consistency and now its more sophisticated but still error prone. One should set the fees based on how fast and accurately they can deliver I suppose. If they made an error that costed you can make an effort to deal with them. There is no excuse for poor quality and low effort even though they have access to Ai. If there is a place to review their work, leave a review and specify the specific type of error. reply | ||
|---|---|---|---|
| silverFork 17 minutes ago │ next [–] There are no moral implications. Assume that all of them will use it and by the way they have used Ai for a decade before this... way before it was even a mature discipline. In some countries its mandatory now. Germany I believe uses some IBM Ai system and few others use Ai where the case loads were piling up. They used call it expert system even years before Ai and it was designed to reduce their work load and improve the consistency and now its more sophisticated but still error prone. One should set the fees based on how fast and accurately they can deliver I suppose. If they made an error that costed you can make an effort to deal with them. There is no excuse for poor quality and low effort even though they have access to Ai. If there is a place to review their work, leave a review and specify the specific type of error. reply | |||
| 15o9o 35 minutes ago │ prev │ next [–] Until the human goes on record with a court as the human lawyer, I'm not sure using AI to produce work product is something we should expect not to happen. Same with doctors. It will take decades for the law and malpractice to catch up with AI. So human lawyer and doctors will become rubber stamps on AI work. The human's job will be to make sure they are using AI best practices (best models, right level of effort etc) and feel ok stamping their name. Reminds me of why Waymo cars still have pedals and steering wheels. So much easier to leave that and get it classified as street legal than fight each city government. The legacy controls are a rubber stamp. reply | 15o9o 35 minutes ago │ prev │ next [–] Until the human goes on record with a court as the human lawyer, I'm not sure using AI to produce work product is something we should expect not to happen. Same with doctors. It will take decades for the law and malpractice to catch up with AI. So human lawyer and doctors will become rubber stamps on AI work. The human's job will be to make sure they are using AI best practices (best models, right level of effort etc) and feel ok stamping their name. Reminds me of why Waymo cars still have pedals and steering wheels. So much easier to leave that and get it classified as street legal than fight each city government. The legacy controls are a rubber stamp. reply | ||
| 15o9o 35 minutes ago │ prev │ next [–] Until the human goes on record with a court as the human lawyer, I'm not sure using AI to produce work product is something we should expect not to happen. Same with doctors. It will take decades for the law and malpractice to catch up with AI. So human lawyer and doctors will become rubber stamps on AI work. The human's job will be to make sure they are using AI best practices (best models, right level of effort etc) and feel ok stamping their name. Reminds me of why Waymo cars still have pedals and steering wheels. So much easier to leave that and get it classified as street legal than fight each city government. The legacy controls are a rubber stamp. reply | |||
| SLiXG 31 minutes ago │ prev │ next [–] ignore AI writing. If a lawyer wrote a very bad quality report, what should the customer do? reply | SLiXG 31 minutes ago │ prev │ next [–] ignore AI writing. If a lawyer wrote a very bad quality report, what should the customer do? reply | ||
| SLiXG 31 minutes ago │ prev │ next [–] ignore AI writing. If a lawyer wrote a very bad quality report, what should the customer do? reply | |||
| AnneTrotter 48 minutes ago │ prev [–] Ironically? Talk to a different lawyer. You might have evidence of legal malpractice. reply | AnneTrotter 48 minutes ago │ prev [–] Ironically? Talk to a different lawyer. You might have evidence of legal malpractice. reply | ||
| AnneTrotter 48 minutes ago │ prev [–] Ironically? Talk to a different lawyer. You might have evidence of legal malpractice. reply |
They used call it expert system even years before Ai and it was designed to reduce their work load and improve the consistency and now its more sophisticated but still error prone. One should set the fees based on how fast and accurately they can deliver I suppose. If they made an error that costed you can make an effort to deal with them. There is no excuse for poor quality and low effort even though they have access to Ai.
If there is a place to review their work, leave a review and specify the specific type of error.
Reminds me of why Waymo cars still have pedals and steering wheels. So much easier to leave that and get it classified as street legal than fight each city government. The legacy controls are a rubber stamp.