# AI is wonderful and will make your life better! (not) **URL:** \ **Category:** The BBQ Pit\ **Tags:** ai\ **Created:** [July 20, 2025, 4:40pm UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117 "2025-07-20T16:40:05Z")\ **Posts on this page:** 20\ **Page:** 6
**Author:** ![kaylasdad99](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/kaylasdad99/32/3398_2.png) [@kaylasdad99](https://boards.straightdope.com/u/kaylasdad99)\ **Post date:** [August 22, 2025, 5:01am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/101 "2025-08-22T05:01:17Z")
I don’t accept the existence of the new _Battlestar Galactica_. Cylons are evil toasters and that’s all there is to it. ---
**Author:** ![Just\_Asking\_Questions](https://avatars.discourse-cdn.com/v4/letter/j/ba9def/32.png) [@Just\_Asking\_Questions](https://boards.straightdope.com/u/Just_Asking_Questions)\ **Post date:** [August 22, 2025, 5:13am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/102 "2025-08-22T05:13:06Z")
> [@kaylasdad99](#): > > I don’t accept the existence of the new _Battlestar Galactica_. Yeah, the old one may have been bad, but it was _our_ bad. I prefer Commander Cartright and my Starbuck to be a guy. I’m old school. John Colicos was a great Baltar. And there’s no Kane like Lloyd Bridges. Imperious Leader can only be Patrick Macnee. If only the plots were better. And the less said about Robbie Rist, the better! ---
**Author:** ![Richard\_Pearse](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/richard_pearse/32/13144_2.png) [@Richard\_Pearse](https://boards.straightdope.com/u/Richard_Pearse)\ **Post date:** [August 22, 2025, 6:04am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/103 "2025-08-22T06:04:51Z")
> [@Just\_Asking\_Questions](#): > > And then there’s **Richard\_Pearse** ’s overly long and frustrating attempt to get an excel file from a pdf via AI, when it seems it should just be easier to…do it Old School. Cut and paste the values into excel. I am very good at spending a whole morning doing something “efficiently” instead of just 5 minutes old school. ---
**Author:** ![Der\_Trihs](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/der_trihs/32/233_2.png) [@Der\_Trihs](https://boards.straightdope.com/u/Der_Trihs)\ **Post date:** [August 22, 2025, 6:32am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/104 "2025-08-22T06:32:57Z")
> [@Just\_Asking\_Questions](#): > > And this is why I fear that we are well on our way to Idiocracy. People will think these are the correct facts, and after a time, they will be, because all us “luddites” will be dead and no one will correct them, and more sadly, no one will care. I doubt it. Either the AI fad will collapse, or the organizations/societies that use it will collapse and be unable to build or maintain them. They are just too incompetent to sustainably run everything they’ve been given power over. There’s “stupid” and there’s “doesn’t work”, and this so-called AI is the latter. ---
**Author:** ![scabpicker](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/scabpicker/32/8268_2.png) [@scabpicker](https://boards.straightdope.com/u/scabpicker)\ **Post date:** [August 22, 2025, 6:56am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/105 "2025-08-22T06:56:39Z")
> [@JohnT](#): > > Interesting, because the OCR type stuff is one of the things I find GPT, et al, to be consistently good at. Well, OCR would come in handy if there was an image in the PDF file, but the content of a PDF is at its heart a bunch of text. In addition to the text visible in the document, it has a bunch of instructions for how to render that text and/or any images embedded in the PDF. If you can interpret that properly and render it correctly, than you can create an image that OCR can take a crack at. But interpreting a PDF can be a trial. Some of them are using the layout language to create the text as _curves_, basically rendering the text in vector graphics. Keeps you from having to include fonts in your PDF, but woah that makes interpreting its text a bear. Interpreting the formatting of columns and making the text within them actually make sense is also a problem, even when the information contained within them is actually stored as text. There are some specialized AI tools for data extraction from PDFs that claim to be as near as perfect as possible, but I’ve never used one. From my experience with other tools, as the document in question becomes more complex, the edge cases creep in. ---
**Author:** ![Atamasama](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/atamasama/32/12961_2.png) [@Atamasama](https://boards.straightdope.com/u/Atamasama)\ **Post date:** [August 22, 2025, 7:24am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/106 "2025-08-22T07:24:17Z")
> [@scabpicker](#): > > Well, OCR would come in handy if there was an image in the PDF file, but the content of a PDF is at its heart a bunch of text. In addition to the text visible in the document, it has a bunch of instructions for how to render that text and/or any images embedded in the PDF. If you can interpret that properly and render it correctly, than you can create an image that OCR can take a crack at. Yeah, it’s like if you’re really good at translating from Latin, but instead of getting Latin you got somebody else’s bad translation from Latin. So now your translation skills aren’t very useful. You’re instead dealing with someone else’s mess. ---
**Author:** ![Just\_Asking\_Questions](https://avatars.discourse-cdn.com/v4/letter/j/ba9def/32.png) [@Just\_Asking\_Questions](https://boards.straightdope.com/u/Just_Asking_Questions)\ **Post date:** [August 22, 2025, 2:55pm UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/107 "2025-08-22T14:55:14Z")
> [@Richard\_Pearse](#): > > I am very good at spending a whole morning doing something “efficiently” instead of just 5 minutes old school. And you reminded me of a recent news article I read about some tech CEO that fired 80% of his staff because they didn’t fully embrace AI. You would have got to keep your job. 🙂 ---
**Author:** ![JohnT](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/johnt/32/15048_2.png) [@JohnT](https://boards.straightdope.com/u/JohnT)\ **Post date:** [August 22, 2025, 2:57pm UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/108 "2025-08-22T14:57:13Z")
> [@scabpicker](#): > > Well, OCR would come in handy if there was an image in the PDF file, but the content of a PDF is at its heart a bunch of text. Well, I’m not “techy”, so “OCR” may not have been the correct wording. Regardless, I upload PDF’s of Franchise Disclosure Documents to Gemini/GPT all the time to extract data and they rarely fail at the task. ---
**Author:** ![Dewey\_Finn](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/dewey_finn/32/4222_2.png) [@Dewey\_Finn](https://boards.straightdope.com/u/Dewey_Finn)\ **Post date:** [August 22, 2025, 3:57pm UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/109 "2025-08-22T15:57:54Z")
> [@scabpicker](#): > > Well, OCR would come in handy if there was an image in the PDF file, but the content of a PDF is at its heart a bunch of text. Only _some_ PDFs contain “a bunch of text.” Plenty are just an image of a scanned document, and that’s where OCR is useful. ---
**Author:** ![scabpicker](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/scabpicker/32/8268_2.png) [@scabpicker](https://boards.straightdope.com/u/scabpicker)\ **Post date:** [August 22, 2025, 4:49pm UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/110 "2025-08-22T16:49:59Z")
> [@JohnT](#): > > I upload PDF’s of Franchise Disclosure Documents to Gemini/GPT all the time to extract data and they rarely fail at the task. Well, yeah. Most PDFs aren’t weird. However, one of the products I support is an email security product that scans literally hundreds of thousands of them a day. The parser doesn’t have a problem with most of them, but every few days or so we get one that it can’t process properly. Sometimes the doc doesn’t actually conform to the PDF spec, sometimes it’s within the spec but uses it in a way that our parser can’t handle. We file a bug on the parser in the latter case. > [@Dewey\_Finn](#): > > Only _some_ PDFs contain “a bunch of text.” Plenty are just an image of a scanned document, and that’s where OCR is useful. The next line in the same paragraph covers that. > [@scabpicker](#): > > In addition to the text visible in the document, it has a bunch of instructions for how to render that text and/or any images embedded in the PDF. Even the ones that only have an image visible still contain a bunch of text describing how to present that image. Not really germane to the conversation: I fucking hate PDFs that are just an image of text. A scan of an old hard copy of a document? Fine, I’ll deal with that, no way around it. But some geniuses will go out of their way to make both our lives harder by taking a screenshot of a PDF that really is nothing but text that you could copy and paste (or a similar document format), and embedding that image into a _new_ PDF. I’m so lucky I’m not close enough to strangle these people. OCR can usually figure it out, but it doesn’t always. Frankly, if it’s an image of a document that I know is actually copy/pasteable text, I won’t even try OCR. I’ll just send them a message saying "Nope not playing your dumb game, here’s how you export that doc in a useful way.’ Fun article from back in March (practically ancient history) about the pitfalls of using LLMs as OCR engines: > **[Why extracting data from PDFs is still a nightmare for data experts](https://arstechnica.com/ai/2025/03/why-extracting-data-from-pdfs-is-still-a-nightmare-for-data-experts/)** > > Countless digital documents hold valuable info, and the AI industry is attempting to set it free. ---
**Author:** ![pirviii](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/pirviii/32/3233_2.png) [@pirviii](https://boards.straightdope.com/u/pirviii)\ **Post date:** [August 23, 2025, 1:20am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/111 "2025-08-23T01:20:43Z")
> [@scabpicker](#): > > I’ll just send them a message saying "Nope not playing your dumb game, here’s how you export that doc in a useful way.’ I wish I were still working so I could do this. ---
**Author:** ![Stranger\_On\_A\_Train](https://avatars.discourse-cdn.com/v4/letter/s/13edae/32.png) [@Stranger\_On\_A\_Train](https://boards.straightdope.com/u/Stranger_On_A_Train)\ **Post date:** [August 23, 2025, 1:28am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/112 "2025-08-23T01:28:35Z")
> **[Elon Musk Warns That AI Is "Obviously Gonna One-Shot the Human Limbic System"](https://futurism.com/elon-musk-limbic-system)** > > The billionaire's latest theory is a contradictory prediction of AI's effects on mental health on one hand, and fertility on the other. > _“That said, I predict — counter-intuitively — that it will increase the birth rate!” he continued without explanation. “Mark my words. Also, we’re gonna program it that way.”_ Stranger ---
**Author:** ![peccavi](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/peccavi/32/388_2.png) [@peccavi](https://boards.straightdope.com/u/peccavi)\ **Post date:** [August 23, 2025, 2:45am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/113 "2025-08-23T02:45:58Z")
Maybe he just got through streaming Demon Seed? > **[Demon Seed](https://en.wikipedia.org/wiki/Demon_Seed)** > > Demon Seed is a 1977 American science-fiction horror film directed by Donald Cammell. It stars Julie Christie and Fritz Weaver. The film was based on the 1973 novel of the same name by Dean Koontz, and concerns the imprisonment and forced impregnation of a woman by an artificially intelligent computer. Gerrit Graham, Berry Kroeger, Lisa Lu and Larry J. Blake also appear in the film, with Robert Vaughn uncredited as the voice of the computer. > Dr. Alex Harris is the developer of Proteus IV, an e... ---
**Author:** ![Stranger\_On\_A\_Train](https://avatars.discourse-cdn.com/v4/letter/s/13edae/32.png) [@Stranger\_On\_A\_Train](https://boards.straightdope.com/u/Stranger_On_A_Train)\ **Post date:** [August 23, 2025, 3:01am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/114 "2025-08-23T03:01:06Z")
I’m pretty sure that this film is Elon Musk’s inspiration, and I don’t mean for artificial intelligence. Stranger ---
**Author:** ![nearwildheaven](https://avatars.discourse-cdn.com/v4/letter/n/90db22/32.png) [@nearwildheaven](https://boards.straightdope.com/u/nearwildheaven)\ **Post date:** [August 23, 2025, 3:03am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/115 "2025-08-23T03:03:02Z")
I heard variations of this when Photoshop was launched in the late 1990s. ---
**Author:** ![scudsucker](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/scudsucker/32/14101_2.png) [@scudsucker](https://boards.straightdope.com/u/scudsucker)\ **Post date:** [August 23, 2025, 3:06am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/116 "2025-08-23T03:06:36Z")
> [@Elmo](#): > > That said, I predict — counter-intuitively — that it will _increase_ the birth rate!” he continued without explanation. “Mark my words. Also, we’re gonna program it that way.” So… How do I calculate pi, ChatGPT? > [@](#): > > Several series converge to π: > > - **Leibniz Formula** (slow convergence): > > π=4(1−13+15−17+19−⋯ )\pi = 4 \left(1 - \tfrac{1}{3} + \tfrac{1}{5} - \tfrac{1}{7} + \tfrac{1}{9} - \cdots \right)π=4(1−31​+51​−71​+91​−⋯) > > I notice your wife is not currently pregnant. Would you like to stop asking math questions and instead I can suggest some romantic or seductive options? ---
**Author:** ![Stranger\_On\_A\_Train](https://avatars.discourse-cdn.com/v4/letter/s/13edae/32.png) [@Stranger\_On\_A\_Train](https://boards.straightdope.com/u/Stranger_On_A_Train)\ **Post date:** [August 23, 2025, 3:07am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/117 "2025-08-23T03:07:51Z")
> [@nearwildheaven](#): > > I heard variations of this when Photoshop was launched in the late 1990s. Did anyone ever evaluate Adobe as having astronomically inflated market valuations or predict that it would put the majority of white collar workers on the dole? > [@scudsucker](#): > > How do I calculate pi, ChatGPT? > > > [@](#): > > > > Several series converge to π: > > > > - **Leibniz Formula** (slow convergence): Only a ponce would use the Leibniz sequence to calculate \pi. Twenty points higher than me, doesn’t know to use a spigot algorithm to calculate transcendental numbers? \*shakes head\* Stranger ---
**Author:** ![Der\_Trihs](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/der_trihs/32/233_2.png) [@Der\_Trihs](https://boards.straightdope.com/u/Der_Trihs)\ **Post date:** [August 23, 2025, 3:16am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/118 "2025-08-23T03:16:19Z")
> [@nearwildheaven](#): > > I heard variations of this when Photoshop was launched in the late 1990s. Nobody tried to put Photoshop in charge of anything, so that analogy doesn’t work. ---
**Author:** ![nearwildheaven](https://avatars.discourse-cdn.com/v4/letter/n/90db22/32.png) [@nearwildheaven](https://boards.straightdope.com/u/nearwildheaven)\ **Post date:** [August 23, 2025, 4:24am UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/119 "2025-08-23T04:24:25Z")
My comparison with Photoshop was WRT how things could be altered, and would most people be able to tell the difference. ---
**Author:** ![wolfpup](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/wolfpup/32/10618_2.png) [@wolfpup](https://boards.straightdope.com/u/wolfpup)\ **Post date:** [August 23, 2025, 5:14pm UTC](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117/120 "2025-08-23T17:14:16Z")
> [@OttoDaFe](#): > > On 7 August OpenAI released GPT-5, which was such a fantabulous leap forward from its predecessors that the company shut off access them. > > It … didn’t go as planned: Deceptive bullshit like this really pisses me off. I’m not saying this didn’t happen, but I’m saying there’s either a great deal of unstated information about what really happened, or some really major glitches in GPT-5, which is a software development issue and not an AI issue. I just asked ChatGPT the exact same question – to name the first 12 US presidents – and the answer to this simple question was predictably unsurprising: > Q: Who were the first 12 US presidents? > > ChatGPT: Here’s the list of the first twelve U.S. presidents in order, with their years in office: > > 1. **George Washington** (1789–1797) > 2. **John Adams** (1797–1801) > 3. **Thomas Jefferson** (1801–1809) > 4. **James Madison** (1809–1817) > 5. **James Monroe** (1817–1825) > 6. **John Quincy Adams** (1825–1829) > 7. **Andrew Jackson** (1829–1837) > 8. **Martin Van Buren** (1837–1841) > 9. **William Henry Harrison** (1841) — served only one month before dying in office > 10. **John Tyler** (1841–1845) > 11. **James K. Polk** (1845–1849) > 12. **Zachary Taylor** (1849–1850) > > Would you like me to also give you a quick note on each (like their main accomplishments or controversies), or just the list is enough? This is why I continue to use ChatGPT\* for a wide variety of genuinely useful functions, within reason and with an understanding of its limitations, while all the uninformed luddites rage about how “useless” and even “dangerous” it is. But the OP is really about a different issue, about the significant potential for abuse of AI by corporations. I completely agree, but it should be recognized that this has been going on since the dawn of the computer revolution when corporations took away large amounts of control and autonomy from their customer-facing agents and turned it over to their computers. Thus in a single move managing to both demoralize their employees and deny them any empowerment and simultaneously piss off their customers. AI will undoubtedly take this to the next level, but fundamentally corporations simultaneously shitting on both rank-and-file employees and customers is nothing new. \* Well, that’s not really why. That’s a trivial example that’s easily Googled. But the information you can get from extended interactive conversations is entirely at a different level. [Previous page](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117.md?page=5) [Next page](https://boards.straightdope.com/t/ai-is-wonderful-and-will-make-your-life-better-not/1021117.md?page=7)