chatGPT

[dw]

I don't find that particularly persuasive. I'm reminded of some of the things people have said about self-driving cars.

Sure they will wind up killing people in ways human drivers would not have...but consider all the people they won't kill that humans would have (e.g. drunk drivers).

Besides interfacing with reality and outputting text are two different things. I'm not sure the difficulties of the one should limit our appreciation of the advances made in the other.

[FredM]

Is this thing ever going to be capable of insight? Will it ever be able to "hear" and critically assess its suggested reharm? Come to think of it, music might be the ultimate Turing test.

No, and probably not anytime in the near future. But I think you underestimate how much of our daily lives, jobs and otherwise are spent doing things that also don't involve insight. Most of what we do is just "work", and this could be pretty revolutionary in eliminating much of that.

Lots of it, sure. But even when I'm on autopilot, there's still some layer of human judgement going on, scanning for unusual dangers and opportunities.
It comes down to how we use it, like any tool. I'm not arguing against literal autopilot on aeroplanes, for example. But AI will be marketed as a replacement for humans, and if people start to believe that hype, it'll be a costly mistake I expect. Aeroplanes should still carry pilots, even when they're not the ones doing most of the flying.

Like I said, this is absolutely going to put a ton of people out of work starting fairly soon (if they don't shut this down). In 5 years the damage is going to be severe. Not because it actually replaces a human being, but because one human being with this tool will be significantly more productive than 10 without it.

And that's being generous. @omaniphil is correct. Knowledge workers have a very inflated view of how much they actually use the "human" part of their brain. From my experience, many (if not the majority) of knowledge workers straight up don't do real work. Realistically they spend 80% of their "billable" time reading internet blogs in their domain at best, and just freaking scrolling through social media at worst. They've been pushing for a decade or more now to turn their jobs into automatable tasks so they can cruise on autopilot and spend more time looking at their phone. That's going to prove to be a costly mistake.

It's going to be really interesting how much it shakes things up long term. We're now in an arms race with an AI in problem decomposition. We're still pretty far ahead today, but as it catches up more and more of the potential knowledge worker jobs are going to get automated away and only the top 1-10% of knowledge workers (as measured in "ability" to decompose problems today) are probably going to have a job in 10 years.

[convergentsum]

I'm still not sure... when I buy a service from a person, sure I want the donkey-work done but what I'm really paying for is for them to give my problem a holistic once-over and alert me to dangers and opportunities. Those alerts may be rare, but we don't want to miss them.
Definitely it will make each worker more productive, and that's disruptive in itself -- see "the man in the white suit" -- I'm just worried about people falling for the hype whole-sale.

[omaniphil]

Like I said, this is absolutely going to put a ton of people out of work starting fairly soon (if they don't shut this down). In 5 years the damage is going to be severe. Not because it actually replaces a human being, but because one human being with this tool will be significantly more productive than 10 without it.

I agree with most of the above, except I think I have a slightly more optimistic take on it. I think whatever turmoil this creates is likely to be short term, and limited in scope, and that we might see huge improvements in overall economic fortune due to the increases in productivity gained. Every time there's been some technological paradigm shift, from the industrial revolution, to manufacturing automation, there are people put out of work, sure, but the increased wealth and prosperity leads to new demands for good/services. I think it is difficult to predict what that might be right now, but I think we'll be ok.

[FredM]

Like I said, this is absolutely going to put a ton of people out of work starting fairly soon (if they don't shut this down). In 5 years the damage is going to be severe. Not because it actually replaces a human being, but because one human being with this tool will be significantly more productive than 10 without it.

I agree with most of the above, except I think I have a slightly more optimistic take on it. I think whatever turmoil this creates is likely to be short term, and limited in scope, and that we might see huge improvements in overall economic fortune due to the increases in productivity gained. Every time there's been some technological paradigm shift, from the industrial revolution, to manufacturing automation, there are people put out of work, sure, but the increased wealth and prosperity leads to new demands for good/services. I think it is difficult to predict what that might be right now, but I think we'll be ok.

Absolutely. People should be excited about this long term. "Damage is going to be severe" is specifically in the context of existing jobs today -- not as a society or general human welfare. The economy isn't a finite pool of resources you divvy up by your chosen economic system. Production is a real thing.

The other side of this is that this is going to incentivize (force) people with human brains to find ways to actually use the human part of those brains. The result is going to be awesome.

I'm still not sure... when I buy a service from a person, sure I want the donkey-work done but what I'm really paying for is for them to give my problem a holistic once-over and alert me to dangers and opportunities. Those alerts may be rare, but we don't want to miss them.
Definitely it will make each worker more productive, and that's disruptive in itself -- see "the man in the white suit" -- I'm just worried about people falling for the hype whole-sale.

I think you're completely missing the point. Your first sentence isn't even true. You 1000% have purchased goods and services without human interaction that would have required human interaction 50 years ago. To use your attempted language here, if the tail risk is MY LIFE, then yeah, I'm going to want a human there. If the tail risk is I lose $100, and not using a human saves me $20, then -- like most people -- I'm probably going to skip the costly human interaction.

You've also 1000% listened to advice from "an AI" if your definition is "no significant human intelligence was used in the creation of this advice." Most of what's on the web today (internet blogs from "experts", news, etc) is generated by an unthinking hive mind -- the exact mechanism of which looks similar to ChatGPT under the hood except the research is shallower and there's no direct oversight (so it's much worse).

[convergentsum]

I'm still not sure... when I buy a service from a person, sure I want the donkey-work done but what I'm really paying for is for them to give my problem a holistic once-over and alert me to dangers and opportunities. Those alerts may be rare, but we don't want to miss them.
Definitely it will make each worker more productive, and that's disruptive in itself -- see "the man in the white suit" -- I'm just worried about people falling for the hype whole-sale.

I think you're completely missing the point. Your first sentence isn't even true. You 1000% have purchased goods and services without human interaction that would have required human interaction 50 years ago.

Of course I have, I was talking specifically about services I've bought from a person -- I had in mind hiring consultants and whatnot.

You've also 1000% listened to advice from "an AI" if your definition is "no significant human intelligence was used in the creation of this advice." Most of what's on the web today (internet blogs from "experts", news, etc) is generated by an unthinking hive mind -- the exact mechanism of which looks similar to ChatGPT under the hood except the research is shallower and there's no direct oversight (so it's much worse).

Yeah, but I doubt I'm confused about what is computer generated and what is presented by a human, and I can bring appropriate expectations to each case... even the case where a human has prepared a shallow summary of some inexpert web searching, I can detect and account for that too. The danger with presenting computer-generated advice as plausibly human-expert-sounding is that we can't immediately tell which is which anymore, so we're likely to read it with inappropriate expectations, at least until we learn the AI "tells".

[FredM]

The danger with presenting computer-generated advice as plausibly human-expert-sounding is that we can't immediately tell which is which anymore, so we're likely to read it with inappropriate expectations, at least until we learn the AI "tells".

Maybe try the product before generalizing conclusions.

ChatGPT in it's current form provides far more nuanced and balanced answers than the vast majority of so called expert humans on the internet. It's pretty telling, that Andrew Ng, who is actively trying to discredit this product because he's working on a competing one, couldn't come up with a single example of what you're claiming except posing the question literally like:

"Describe overconfidence in the tone of someone who is themselves overconfident."

Hysterically he posted that interaction with ChatGPT as evidence of what you're saying.

In reality, every question of mine it answers it caveats with warnings and nuance. When I asked it to create user stories and estimate them it led with a one paragraph explanation on why software estimates are difficult and how its answer shouldn't replace those made by a real human. It concluded it's answer with a reminder of that and some nuance around its estimates.

It provided my training plan, but followed it up with:

"Note: It's important to listen to your body and adjust the volume and intensity of your workouts as needed. Make sure to warm up properly before each workout and cool down afterwards with stretches. It's also important to progressively increase the weight you are lifting over time to continue challenging your muscles and improving your strength."

Again -- this is more useful than many personal coaches claiming expertise in this "field" (SS -- you're a pu**y anyone?)

It's not an accident that 1/4 of OpenAI's job postings are for their "Trust and Safety" division. Greg B. and team have put an insane amount of thought into providing an AI that does the opposite of what you're claiming. From what I can tell they've succeeded. Hugely. I'd probably rather listen to ChatGPT then most "experts" I've read opinions on.

[omaniphil]

While we're on the topic of insight from a person vs AI hype, look at this summary of the Iran Iraq war it generated for in iambic pentameter. This is better than I could have done.

[FredM]

@omaniphil

yeah... like I said... 5 years ahead of anything I thought possible just 2-3 weeks ago.

My CEO has already used it to generate blog content and is basically pasting it as is.

Our lawyer has already used it to generate a new terms of service given some constraints and likewise is basically using it as is.

Our sales lead used it to generate a marketing plan leading up to a conference and declared it's better than the one he spent days on and is using it almost as is.

The breadth of its ability is more impressive to me than the depth -- although the depth, again, is still way ahead than what I thought was possible in 2022.

One of the cooler things I've seen online is using it to generate a meal plan. I'm going to try that soonish -- but its ability to fairly accurate plan meals given both calorie and available food (or preferred food) constraints is a game changer. It's one of the least impressive things it can do but one of the cooler use cases lol.

[5hout]

Draft the 8 different main jokes to use in a stand-up comedian's 5 minute working set based on relationships between people of different socioeconomic. The stand-up comedian's working set should be grounded in inner city washington DC life and discuss differences from that and an upper class upbringing in a rich area. Do not include opening or closing material.
Here are eight main jokes for a stand-up comedian's set based on relationships between people of different socioeconomic backgrounds in inner city Washington D.C. and an upper class upbringing:

"When I first met my husband, I was like, 'Whoa, where did you learn to speak English so properly?' And he was like, 'Uh, it's called an education.' And I was like, 'Oh, right. I've heard of that.'"

"Like, when I invited him over to my family's house for Thanksgiving, he was shocked to see my uncle grilling a turkey on the front porch. He was like, 'I've never seen anyone cook a turkey outside before.' And I was like, 'Yeah, it's called a tradition.'"

"And don't even get me started on the grocery store. My husband was used to going to these fancy supermarkets with everything you could possibly imagine. But when we went to the corner store in my neighborhood, he was like, 'Where's the organic kale? The gluten-free quinoa? The artisanal cheese selection?' And I was like, 'Dude, we're lucky if they have toilet paper.'"

"My husband's family is used to going on these fancy vacations to exotic places like the Caribbean or Europe. But when I took him to visit my family in rural Virginia, he was like, 'Where's the beach? The museums? The fancy restaurants?' And I was like, 'Dude, we're lucky if we have electricity.'"

"My husband's family is used to eating these fancy meals with multiple courses and fine china. But when I took him to my grandma's house for dinner, he was like, 'Where's the soup course? The salad course? The dessert course?' And I was like, 'Dude, we're lucky if we have a main course.'"

"My husband's family is used to living in these fancy houses with multiple floors and a pool. But when I took him to my apartment in the city, he was like, 'Where's the yard? The garage? The guest room?' And I was like, 'Dude, we're lucky if we have a living room.'"

"My husband's family is used to having these fancy cars with all the bells and whistles. But when I took him to see my cousin's car, he was like, 'Where's the navigation system? The sunroof? The heated seats?' And I was like, 'Dude, we're lucky if it starts.'"

"My husband's family is used to having these fancy jobs with high salaries and benefits. But when I took him to meet my friends at their jobs, he was like, 'Where's the office? The desk? The computer?' And I was like, 'Dude, we're lucky if we have a chair.'"

This took 1.5 hours of, but most of that was spent trying to make it be Dave Chapelle.

ETA:

The 2nd and 3rd jokes are the best accidental commentary on assimilation:

Draft the 3 different main jokes to use in a stand-up comedian's 5 minute working set based on being a Jewish person living in an Anglo-Saxon protestant society and observations on life in the city Do not include opening or closing material. Here are three main jokes for a stand-up comedian's set based on being a Jewish person living in an Anglo-Saxon Protestant society and observations on life in the city:

"I love living in this city, but sometimes I feel like I stick out like a sore thumb. Like, I was at this fancy cocktail party and someone asked me where I'm from. I said, 'I'm Jewish.' And they were like, 'Oh, is that a country?' And I was like, 'No, it's a religion. You know, like Christianity or Islam.' And they were like, 'Oh, right. Like, Buddhism or Hinduism?' I was like, 'Uh, I'm pretty sure that's not the same thing.'"

"I love the diversity of this city, but sometimes it can be a little overwhelming. Like, I was at the park and I saw this guy wearing a yarmulke. I was like, 'Whoa, what's the deal with that?' And he was like, 'Oh, it's a traditional Jewish head covering.' And I was like, 'Uh, I'm pretty sure that's not traditional for everyone.' And he was like, 'Well, it is for me. It's a reminder of my faith and heritage.' I was like, 'Uh, I'm pretty sure that's not the same thing either.'"

"I love the culture of this city, but sometimes it can be a little confusing. Like, I was at this museum and I saw this painting of a guy with a long beard and big nose. I was like, 'Whoa, what's the deal with that guy?' And the tour guide was like, 'Oh, that's a portrait of Moses. He was a prophet and leader of the Israelites.' And I was like, 'Uh, I'm pretty sure that's not what makes someone a prophet.' And the tour guide was like, 'Well, in his time, a long beard and big nose were considered signs of wisdom and authority.' I was like, 'Uh, I'm pretty sure that's not the same thing either.'"

If you really generously read this a comment on assimilation, and not on confused garble, it's pretty decent. I also tried a variation that got a little spicy, but lost the actually sort of funny assimilation subtext and replaced it with weird anti-Irish sentiment:

Here are three main jokes for a stand-up comedian's set based on being an Ashkenazi person living in an Anglo-Saxon Protestant society and observations on life in the city:

"I love living in this city, but sometimes I feel like I stick out like a sore thumb. Like, I was at this fancy cocktail party and someone asked me where I'm from. I said, 'I'm Ashkenazi.' And they were like, 'Oh, is that a country?' And I was like, 'No, it's a genetic marker for Jews of Eastern European descent. You know, like a blue eye for blondes.' And they were like, 'Oh, right. Like, a club foot for the Irish?' I was like, 'Uh, I'm not sure that's the same thing.'"

"I love the diversity of this city, but sometimes it can be a little overwhelming. Like, I was at the park and I saw this guy walking a pet alligator on a leash. I was like, 'Whoa, where did you get that?' And he was like, 'Oh, I bought it at the exotic pet store.' And I was like, 'Uh, I'm pretty sure that's illegal.' And he was like, 'Well, I don't know about that. But I do know it's cheaper than a dog.' I was like, 'Uh, I'm pretty sure that's not the same thing either.'"

"I love the culture of this city, but sometimes it can be a little confusing. Like, I was at this museum and I saw this painting of a guy with an enormous head. I was like, 'Whoa, what's the deal with that guy?' And the tour guide was like, 'Oh, that's a portrait of Sir Isaac Newton. He was a brilliant scientist and mathematician.' And I was like, 'Uh, I'm pretty sure that's not what makes someone brilliant.' And the tour guide was like, 'Well, in his time, a large head was considered a sign of intelligence.' I was like, 'Uh, I'm pretty sure that's not the same thing either.'"

Three more:

Draft the 3 different main jokes to use in a stand-up comedian's 5 minute working set based on having a typical ashkenzai mother married relaxed lapsed catholic farther. Do not include opening or closing material. "My mom is the quintessential Ashkenazi mother - she's always asking me when I'm going to get married and give her grandchildren. And my dad is a lapsed Catholic, so he's like, 'Just do whatever makes your mother happy.' So I'm like, 'Great, I'll just convert to Judaism and marry a nice Jewish boy. That'll make everyone happy!'

"My mom is always trying to set me up with 'nice Jewish boys.' And my dad is a relaxed, laid-back guy, so he's like, 'Hey, just go with the flow. If it works out, it works out.' But I'm like, 'What if I want to marry a nice Muslim boy? Or a nice Hindu boy? Or a nice atheist boy?' My mom's response is always the same: 'As long as he's Jewish.'

"Growing up with a typical Ashkenazi mother and a lapsed Catholic father was interesting, to say the least. My mom would always drag us to synagogue on the High Holidays, and my dad would be like, 'Hey, at least we get a day off from work!' But seriously, it was a nice balance - my mom kept us grounded in our cultural traditions, and my dad taught us to be open-minded and accepting of other religions and beliefs.

[51M0N]

So it's good for finding cocktail recipes. On the other hand it totally failed a coffee question that I thought it would easily get from info on the internet "what is a Cafelat Robot?"

The issue being that as somebody who is into coffee I know how absurd its response is when it says it has a 15 bar pump and steam wand (it's a manual lever machine), but for somebody who doesn't know anything about coffee it sounds like a reasonable response.

[xuerebx]

I tried asking it some specific technical accounting questions and it gets things wrong most of the time. This is great news, because the moment it starts spitting good results I lose my job

[JimRiley]

Q: This is Worker speaking. Hello!

A: Hello Worker! It's nice to meet you. How can I help you today? Do you have any questions or are you looking for information on a specific topic? I'm here to assist you and provide information on a wide range of topics, so please don't hesitate to ask if you have any questions or need help with anything.

Q: Oh, you don't know about "I Think We're All Bozos On This Bus"? Siri does.

A: I'm sorry, but I am not familiar with a specific reference to a book, movie, or other work titled "I Think We're All Bozos On This Bus." As a language model, I do not have the ability to browse the internet or access new information beyond my training data, which was cut off in 2021. Without more context or details, it is difficult for me to provide any further assistance or information on this topic.

If you have any specific questions or are looking for information on a particular topic, please let me know and I will do my best to help.

Q: Why does the porridge bird lay his egg in the air?

A: The porridge bird is a fictional creature that appears in a well-known children's nursery rhyme:

"The porridge bird lays her egg in the air,
The egg falls down, nobody knows where;
But when it breaks, there's a lot of porridge everywhere."

The rhyme does not explain why the porridge bird lays its egg in the air, and there is no scientific or factual basis for the behavior described in the rhyme. The porridge bird is a purely fictional character that is used to entertain and amuse children, and the rhyme should not be taken literally.


Well, that was no fun...

[augeleven]

While we're on the topic of insight from a person vs AI hype, look at this summary of the Iran Iraq war it generated for in iambic pentameter. This is better than I could have done.

It’s not very daDUM daDUM daDUM daDUM daDUM though.

[Culican]

Lawyer apologizes for fake court citations from ChatGPT

[aurelius]

Lawyer apologizes for fake court citations from ChatGPT

The only surprising thing is the Court caught it. Let's just say in the lower level courts lawyers and judges play fast and loose with the law.

[Culican]

I asked ChatGPT for the proper FDA Food Code sections to use when citing a violation in a food establishment.

It gave a very plausible sounding answer with a code section and the wording in the section that applied to the situation I had described. Unfortunately, the code section it cited has nothing to do with the scenario I described and the wording it gave me does not exist in that section. When I told it that, it apologized and said that there is nothing in the code addressing the situation. When I gave it sections that do apply, it apologized again.

For those who read this, think about how many times you can hear people watching videos or talking on their phone while they are in a public restroom taking a crap.

The scenario below was observed earlier today in a hospital kitchen. Here the text of my interaction:

Me: Consider the following scenario: A sanitarian is inspecting a kitchen for a local jurisdiction that uses the 2017 FDA Food Code. The sanitarian observes a food employee cooking food on a grill. The food, when done, will be placed in a pan on a steam table about 5 feet away from the grill. The sanitarian further observes a smartphone, which is playing a news broadcast, sitting on the edge of the steam table approximately 2 inches from the pan that will be used for the cooked food.

ChatGPT: Under the 2017 FDA Food Code, the specific section that would likely be applicable to the situation described in item 3 is Section 3-301.11, which addresses the separation of personal items from food and food-contact surfaces. This section states:

"Except as specified in paragraphs (B) and (C) of this section, personal items such as employee clothing, purses, and cell phones shall be stored in designated areas and shall not be allowed to cross over or touch clean equipment, utensils, and single-use articles."

The purpose of this section is to prevent the potential for contamination of food, utensils, and equipment by personal items, which can harbor bacteria or other contaminants. Based on the scenario you provided, the smartphone playing the news broadcast should be stored in a designated area away from food and food-contact surfaces to comply with this section of the FDA Food Code.

Me: I can find the wording you cite nowhere in Section 3-301.11. This section applies to bare hand contact with ready to eat foods. It does not mention the storage of employee possessions.

ChatGPT: Apologies for the confusion caused. You are correct that Section 3-301.11 of the 2017 FDA Food Code does not specifically address the storage of personal items. I apologize for the incorrect reference.

Regarding the storage of personal items, the FDA Food Code does not have a specific section dedicated to it. However, many local health departments and jurisdictions may have their own regulations or guidelines regarding the storage of personal items in food preparation areas. These regulations are typically in place to prevent potential contamination of food and maintain a clean and sanitary environment.

Me: Actually sections 6-305.11 and 6-501.110 seem like they do address the storage of personal items.

ChatGPT: Apologies for the oversight. You are correct, and I apologize for the confusion in my previous responses. Sections 6-305.11 and 6-501.110 of the 2017 FDA Food Code do address the storage of personal items in food establishments. Here's a brief overview of these sections:

Section 6-305.11: Storage of Employee Personal Belongings:
This section states that employee personal belongings such as clothing or other items shall be stored in designated areas separate from food, utensils, and single-use items. It aims to prevent cross-contamination between personal items and food preparation areas. The specific requirements regarding the designated storage areas may vary based on local jurisdiction regulations.

Section 6-501.110: Personal Cleanliness:
This section covers the personal cleanliness of food employees, including the storage of employee personal items. It states that food employees must store personal items in a designated area to prevent contamination of food, utensils, equipment, and other items that come into contact with food.

It's important to note that while these sections of the FDA Food Code provide general guidance, the specific requirements and interpretations may vary depending on the local jurisdiction. Consulting the local health department or the specific regulations applicable in your area will provide the most accurate information regarding the storage requirements for personal items in food establishments.

Me: Actually, I work for a local health department. Given your errors, I don't think any of our human inspectors need to fear for their jobs yet. Thank you for your time though.

ChatGPT: I apologize for any confusion or misinformation caused by my previous responses. As an AI language model, I strive to provide accurate and helpful information, but I can make mistakes. I appreciate your understanding, and I'm glad to hear that human inspectors are still essential in ensuring food safety.

[51M0N]

The more I use ChatGPT the less worried I am that it's going to take my job. It's a pretty good copy-editor, but for anything that requires specific information it's not trustworthy. It's prone to hallucinations (i.e. making up info), and since it can't possibly cite its own sources, you effectively need to line by line fact test it.

[aurelius]

The more I use ChatGPT the less worried I am that it's going to take my job. It's a pretty good copy-editor, but for anything that requires specific information it's not trustworthy. It's prone to hallucinations (i.e. making up info), and since it can't possibly cite its own sources, you effectively need to line by line fact test it.

That is easily correctable. The free online version simply references publicly available sources. Understanbly, ChatGPT has trouble sifting through the torrent of bullshit that is the internet. So do people. But you can download an offline version and provide it with only the materials you want it to reference. We are doing this currently with various government manuals and codes. This is very useful as code addressing specific items are spread over separate and multiple documents. You only checked the stormwater critera for channel requirements...didn't think to check the building code or roadway manual? ChatGPT can near instantly search through all of these references and pull out the relevant material.

[Culican]

The more I use ChatGPT the less worried I am that it's going to take my job. It's a pretty good copy-editor, but for anything that requires specific information it's not trustworthy. It's prone to hallucinations (i.e. making up info), and since it can't possibly cite its own sources, you effectively need to line by line fact test it.

That is easily correctable. The free online version simply references publicly available sources. Understanbly, ChatGPT has trouble sifting through the torrent of bullshit that is the internet. So do people. But you can download an offline version and provide it with only the materials you want it to reference. We are doing this currently with various government manuals and codes. This is very useful as code addressing specific items are spread over separate and multiple documents. You only checked the stormwater critera for channel requirements...didn't think to check the building code or roadway manual? ChatGPT can near instantly search through all of these references and pull out the relevant material.

Obviously I used the public version. Where do you think it got all of the erroneous information regarding the FDA Food Code given that the code is readily available at fda.gov?