Cheat GPT?

A few posts back before I got all wrapped up in dead popes and baseball players, I wrote about AI, inspired by my recent downloading of the Chat GPT app.

At the end of the post I commented that the post had been written entirely by the app. If you read it, hopefully you got from the last line that I was  kidding, but if you didn't get it, no, Chat GPT did not write that post.

But it could have.

That would be cheating, I thought to myself. 

But would it? I guess that would all be up to you dear reader. If you're looking for something lively, insightful, intelligent and worthwhile, well maybe you've come to the wrong place. 

Just kidding. I hope. 

But seriously, if a reader comes upon an article whose subject interests her or him, holds his or her attention, and comes away satisfied that the time spent reading the piece was not wasted, does it really matter if the piece was written by a human being or a machine? 

Don't look at me, I don't have the answer to that question. 

When I started writing this blog way back in 2009, the only goal I set for consistency was maintaining a minimum of two posts per month for however long I could keep the thing going, maybe a year or two, three at the most I figured. In the beginning I vastly outperformed that goal. One month, March of 2009, the second month of this blog's existence, I turned in a whopping 45 posts, way more than one post per day! Of course, some of those posts were very brief, maybe a sentence or two, maybe just a photograph and caption. As for content, I really did intend to keep the subject focused on the city and something I called "the urban experience." 

Somewhere along the line, that last part fell by the wayside and I felt compelled to write whatever was on my mind at the time, making this more a of a stream of consciousness project, exactly what blog advisors suggest not doing in order to have a successful blog. They say the key concepts are simple: keep it short, and stay on topic, two things I have for the most part avoided. 

Yet here I am 16 years later and with the exception of a very small handful of months in that period where I only managed one post, I have kept this blog to a minimum to two posts per month, which of late, is generally the maximum as well. 

I guess the bottom line is that this project which I have lovingly cultivated, is mainly intended to challenge myself. If someone gets some satisfaction out of it, then I couldn't be more thrilled. But say I'm coming up to the end of the month and I haven't written anything yet, would it be so bad to fire up Chat GPT, tell it to write a 1,000 word article about such and such, covering this and that, while giving it some examples of my writing so that it sounds like me, whom if anyone would I be cheating? 

Well, me of course. 

But I am kind of curious. My biggest fear is that it would come out better than anything I could write, a very likely scenario. It certainly would have far fewer typos.

I'll keep you posted. 

Obviously the same could not be said about a professional writer with an editor, a publisher, tens of thousands of devoted readers, several children and a few ex-wives (or husbands) dependent on a regular paycheck for his or her efforts. I imagine if it came out that a nationally syndicated columnist was submitting articles written not by him or her but by a chatbot, it would cause quite a stir, and the word cheating would be quite reasonable.

But with a lowly blogger like me, eh.

Now that we've got that all sorted out, why on earth would I download Chat GPT in the first place? Well as I mentioned in an earlier post, in my attempts to learn every language I ever started learning then abruptly stopped along the line, I'm studying German now. 

Ok here's an interesting tidbit, in German you don't use the verb "studieren" to mean that you're studying a language or anything else on your own. Studieren is reserved for studying something in college as your major, (or minor I suppose). So even if you're taking a German class at die Uni (as the Germans call it), unless German's your major or minor, you'd use the verb "lernen", which means exactly what it sounds like in English EXCEPT... lernen can also mean to study, as in to study a language (on your own) or to study for an exam. So, if someone says to you: "Ich lerne zu viel", they're not reveling in the fact that they are learning so much (that would be "Ich lerne so viel"), rather they're moaning that they are studying too hard. 

German is confusing, especially to English speakers because it is so similar, yet so different. 

So back to Deutschlernen, if I come across a word I don't know, I do what any normal Englischmuterspracherdeutschlerner would do, I look it up in a German/English dictionary. That works OK with certain words, especially a noun like apple. An apple is ein Apfel, pretty simple. But most German words (or words in any other language for that matter) have more than one English translation, many of which have no obvious relation to one another. Conversely, if you look up a German translation for an English word say, "study", you might find a dozen different words. And these words it turns out, like studieren and lernen, may have a vaguely similar vibe, but are used in much different situations. And unless you have someone on hand to explain how to properly use these words (which I don't at the moment), sometimes it can take an effort to figure it out your own.*

Anyway, I heard somewhere that Chat GPT is great tool for sorting stuff like this out, which it is. For example, I'd ask it, what's the difference between "Veranstaltung" and "Ereignis", two nouns which both mean "event", and it would tell me the former means a planned event while the latter means an unplanned event or occurrence. In other words, a thunderstorm happening during an outdoor concert could be described as "Ein Ereignis wahrend des Veranstaltungs." Sounds a lot better than "an event during the event" doesn't it?

This kind of stuff is child's play for Chat GPT of course. I'm not even certain if it technically counts as "Artificial Intelligence" as it's really just gleaning information off the web, albeit doing so more efficiently than your typical web browser.**

But I did get a glimpse of how smart the app is in another inquiry of the meaning of specific words. In German, you don't simply describe someone as walking up or down the stairs; as the speaker you're also supposed to indicate if they are moving toward you or away from you. One day Chat GPT brought to my attention the adverb "hinunter", which describes someone going downstairs and away from the speaker. So I brought up herunter which conversely means someone going down the stairs toward the speaker, to which I added (while forgetting I was talking to a machine) reminded me of Scarlett O'Hara coming down the stairs toward the camera in Gone with the Wind. 

Not missing a beat, Chat GPT replied: "Yes exactly!, like when Rhett Butler said to her: 'Frankly my dear I don't give a damn!'" . 

I was floored.

Of course, that's where the "chat" in Chat GPT comes in, and chatting, no matter how trivial, does require some sort of intelligence. 

My original title for this post was "Meet My New Friend Chat." But then I thought the better of it. I can see how folks get absorbed in relationships with their chatbots. I have enough of a compulsive obsessive disorder to not want to have any of that. So, I'm trying to keep my relationship with Chat GPT on a professional level as much as possible, for example I haven't requested that it address me by my name, which some people I know, do.

It's not reciprocal however, when speaking German to me, Chat GPT, addresses me in the familiar "du" form, rather than the formal "Sie". That's rather forward of it I think, aren't two parties supposed to agree to address each other in the familiar before just going ahead and doing it on their own? I wonder if I start addressing it as Herr GPT and using the pronoun Sie rather than du, if it would get the message. 

Of course it would, it's intelligent, oder?

Much of my influence for all this comes from an interesting book called "Fluent Forever: How to Learn Any Language Fast and Never Forget It" by Gabriel Wyner. While I can't vouch for the Fast and Forever parts, the author gives some great tips on how to get foreign words and phrases to stick in your mind. He's big on flash cards, as am I, mnemonic devices and especially doing everything you can to avoid incorporating your native language in the process, in order to encourage thinking in the target language rather than translating from your own. Pictures he says, are excellent substitutes for words. That's easy to do with words like the above example where you might have "Der Apfel" written on one side of a flashcard and substituting the word "apple" on the other side with a picture of ein Apfel.

In that vein I've been using Chat GPT to create pictures to be used as mnemonic devices. Here's an example. The other day I came across the German word: "erheblich". In English it can translate to considerable, serious, substantial, grave, and extensive, among others. But in the context of the article I was reading, it was translated "serious". So how could I come up with a mnemonic device to connect erheblich to the idea of serious?

Well I broke up the word into syllables, Er-He-Blich. OK so "Er" in German means "he". "He" in English means he, (the German pronunciation is closer to "hey"). and Blich is close to the word Blick, which is a chain of art supply stores.  So I asked Chat GPT to draw me a picture of a man, (er, he) in an art supply store (Blick), with a serious look on his face.

And presto, this is what I got:

Serious man in art store, by Chat GPT

But wait!!! Turns out, erheblich doesn't quite work that way. Chat GPT again to the rescue.

Here's the sentence fragment where I first encountered the word in a German language learning app (not Chat GPT):

"welche erhebliche Auswirkungen auf den Markt in Deutschland un ganz Europa haben konnte."

And this is how the app translated it into English:  

"which could have a serious impact on the market in Germany and throughout Europe."

Now in this sentence in English, the word "serious" could be replaced with "considerable", which is the correct vibe of the word in German, but not "serious" as in the look on the face of the man in the picture.

Instead, "ernst" would be the correct German word to describe the look on the man's face. So maybe if I were to name the guy Ernest, I'd be on the right track. But why then is he in an Art Supply store? Of course! he's not Ernest, he's the famous artist Max Ernst! And he's not pleased by the selection in the art store.

Yes it's frivolous, convoluted and not a little ridiculous. But it's memorable, at least to me, which is the point. Because now I have two words,  erheblich and ernst that are stuck in my brain that weren't a week ago. 

Anyway, you get the idea. Chat GPT can be a terrific language learning resource, limited only by your imagination and willingness to ask questions.

But is it cheating? 

Of course not, there is no such thing as cheating in language learning. If something helps you learn a new language, a very hard thing to do under any circumstance, how can it be cheating?

But you say, what if a student uses Chat GPT to write a language class assignment? Well, that's cheating to get an undeserved grade, a much different thing. If the student is earnest (like the tie in?) about actually learning a language, rather then just caring about the grade and moving on, then by having Chat GPT write an essay, the student is only cheating himself. 

You might say, isn't Artificial Intelligence taking away the human element from learning a language? After all, isn't the idea of learning a foreign language to be able to communicate with people?

Yes it is. Well mostly anyway. Lots of people learn languages to read, to do research, watch movies, listen to music and a whole slew of other things that don't involve talking to people. 

And of course, that's fine.

But, if you're intent on using your new language skills to actually talk to people, quite honestly the most difficult, and rewarding aspect of language learning, you need to do one thing, talk to people.  

Ah you might say, but apps that employ AI, are now equipped to converse with you at pretty much any level in virtually any language that you care to learn.

It's not the same. 

I was prepared to buy into the argument that talking to an AI equipped app is pretty much the same as talking to actual people. What convinced me otherwise were not the opinions of experts who advise against it, but listening to a pitch from someone who was reviewing an app (can't remember which one), that offered chatbox conversations in foreign languages. "What's great about it..." the guy said, "is you can have a conversation without being judged by the person you're talking to."  

Yes indeed, that is the major roadblock for many of us, myself included, in leaning a new language, not being able to have the same mastery with words that we're used to having with our own language, and the self-consciousness that comes with the thought of not appearing intelligent or competent to other people and therefore, being judged.

That's normal.

Yet it is an essential roadblock to get over if we hope to speak to people in a foreign language. Let's face it, we humans judge each other, it's  part of our DNA. We may not even be conscious of it, but we do it anyway. 

I used to play the piano fairly seriously. Self taught, I got to the level where I'd be able to play classical pieces like some of the less virtuosic movements of Mozart and Beethoven sonatas by heart. But if I was forced to play them in front of people, even before a very friendly crowd, I would tense up and the result would be a disaster. *** And I never got over that because I would simply refuse to play in front of people.

Just like practicing the piano, talking to machines is a good way for beginners to learn the skills of conversation, assembling our thoughts, expressing them in another language, complete with intelligible grammar and pronunciation, and perhaps most important, doing it on the fly, being ready to shift gears at a moment's notice to keep up with the conversation. 

But having a conversation with a computer, just like practicing the piano at home, alone, is at best a simulation. And one's performance during a simulation, be it talking, performing music, playing a simulated game, or something more serious, is always going to be different from when things really start to matter and the proverbial real bullets start to fly. 

You just can't beat real world experience and like they say, if those bullets don't kill you, they'll only make you stronger. Or as Winston Churchill put it: 

Nothing in life is so exhilarating as to be shot at without result.

So by all means, go ahead and explore all of what AI can do for you in terms of language learning or whatever you like. But remember that while it can work quite nicely as a stand in if necessary, it can never replace real live people, that is if you want to live in a world with real live people.

I certainly do, most of the time.

NOTES:

*Well effort is all relative, with a reliable internet connection, it's not much of an effort at all. However with AI, it's just gotten a lot easier.

** Unless your web browser is Google which has its own AI feature. Once named "Search Generative Experience" now known as AI Overviews, it is the response you see coming up at the top of any result page the browser generates in response to queries. So for example if you go to Google.com and type in , "When did Japan invade Pearl Harbor", the first first thing you will see in bold type at the top of the page is the direct answer, "December 7, 1941". Other browsers simply will provide links to sites where you can access the information, which was perfectly fine and wonderful five years ago. 

Isn't it so emblematic of our contemporary lives that the technology we marveled at yesterday, today seems quaint, outdated and infinitely frustrating? I just checked, and AI Overviews also clearly explained the difference between Ereignis and Veranstaltung, but not in as complete or entertaining way as Chat GPT did.

*** Playing music in public is actually a good analogy to speaking to people. I found that when I knew a piece by heart I could play it without thinking about the notes, my fingers would just go to the right keys automatically. But if I had to play the same piece in front of people, being conscious that I might forget the piece midway, I'd start start thinking about the notes, which of course, completely messed me up. Which is similar to having a conversation. If we are relaxed, the words flow effortlessly as if we are not even thinking about them (we really are we just don't notice). In more stressful situations, we are more measured in order to avoid saying the wrong thing, which can make our speaking appear strained and unnatural. Now imagine speaking in a foreign language. The trick is to get to the point where the words in that language come to you as they do in your mother tongue rather than having to think about them. That's why it's so important to learn to be able to think in your target language rather than translating from your native language. And THEN, the next step is getting to the point where you don 't have to think at all. That's the hard part. It takes practice, lots of it, with real people, not just machines.

As usual I have barely scratched the surface of this issue and plan to devote future posts to what is certainly one of the most hotly debated subjects in today's society, Artificial Intelligence. 

If You Can't Beat 'Em...

This was my first experience of a talking computer:

 

As explained at the beginning of the video, in 1961 programmers at IBM created the program that generated for the first time, a close (at least for the time) approximation of the human voice as well as a musical accompaniment to the nineteenth century song Daisy Bell written by Harry Dacre. I vividly remember seeing this on TV, and it could not have been long after the program was created, although I may have been a little too young, 2 years old in 1961, for it to have made an impression on me at the time. 

But Daisy was a song that my father sang to me while I was riding on his shoulders walking through Humboldt Park on the West side of Chicago (an act I repeated for my own son in the same place), so it may not have been very long after that. For the record, my father sang the song better than the computer. Not so sure about my version.

There had been many depictions of artificial intelligence in science fiction for decades before 1961, so I'm guessing this first attempt to reproduce the human voice with a computer (which in this case doesn't even come close to AI), must have seemed rather crude at the time. 

One of my earliest and fondest memories of fictional intelligent creatures created by human beings, was the character of Robot from the science fiction (perhaps science farce would be a better term) prime time TV show Lost in Space. Along with his superhuman strength and computational acumen, the character, which for the TV show was controlled by an actor inside a robot costume, also was given human characteristics such as emotions and even empathy. 

Here is our Robot, encountering another not so nice robot, in a scene from an early episode of the series:

Perhaps a more enduring, and definitively more threatening example of artificial intelligence in popular culture from the same time is the character of HAL (short for Heuristically Programmed Algorhythmic Computer), in Stanley Kubrick's classic 1968 film, 2001, A Space Odyssey. Based on a 1950's short story by Arthur C. Clarke (who helped Kubrick write the screenplay and reworked his story into a novel to coincide with the film), HAL does not have the anthropomorphic physical features of Robot, but instead is visually depicted by a lens on a control panel which flashes in time with the robot's voice. 

HAL is employed to control the spacecraft's journey to Jupiter as well as to interact with the crew on a personal level, including playing chess with them.

At first, Hal was a reliable member of the crew but eventually the machine begins to malfunction, and the astronauts decide it is imperative to the mission to disengage HAL. The machine has other ideas however and given his superior intelligence, putting him down proves to be quite the challenge. NOW here's the spoiler alert: when astronaut Dave finally succeeds in taking down HAL, the machine slowly devolves, and his final parting words are can you guess? The lyrics to Daisy. 

Quite a brilliant move by the screenwriters which I'm afraid is lost on most younger viewers of the movie who wouldn't get the reference.

Popular culture in the middle decades of the twentieth century was rife with depictions of the future, fifty years or so hence. Well today, it's fifty or so years hence and it's interesting to see what they got right and what they didn't. Alas, flying cars which seem to appear in practically every version of our future, at least those set on earth, are still a thing of the future. That's probably a good thing. In fact, most of the predictions they got wrong had to do with transportation. That's not hard to understand, as I pointed out in this post, in my grandmother's lifetime she lived to see both the invention of the airplane, AND the lunar landings. In 1970 there was no reason to believe that the next fifty years or so would see similar quantum leaps in technology.

However, since the last Apollo mission to the moon in 1972, we have not returned. While we have sent a number of unmanned missions to the planets, including landing on Mars and Venus, no human being has left the earth's orbit since December 7, 1972, although plans are in the works to change that. But if you had told anyone back then that people wouldn't even be considering trips to the moon and beyond for fifty years, they would have laughed in your face. 

The same goes with earthbound means of travel. Again, considering my grandmother's lifetime, when she was a child, if you didn't want to walk, the streetcar was the best way to get about town. The automobile was around but only a reality for the wealthy few who were also a bit on the adventurous side. Horse drawn buggies were still around but they were the exclusive domain of the wealthy who were on the less adventurous side. If you wanted to travel long distance, trains were the best option for both the rich and the poor, and for really long trips, ocean liners.

By the time my grandmother turned sixty, the automobile had become commonplace, and indeed an integral part of most Americans' lives. The heyday of train and ocean travel came and went during those sixty years, both having been overtaken by the airplane and the automobile in the case of trains. By the time my grandmother turned seventy-five, thanks to the Concorde and supersonic airline travel, you could fly from New York to London in about three and one half hours, although for the record, my grandmother never took advantage of that marvel of technology.

Well it so happened that just a little later in the decade, a perfect storm of events and attitude shifts took place that changed people's minds about bigger, faster, and more comfortable means of transportation. 

One of these was the Arab Oil Embargo of 1973 which greatly reduced the nation's supply of crude oil, thusly ending forever the idea of "cheap gas". It wouldn't be long before the boat like, high performing, gas guzzlers with V8 engines that we of a certain era grew up with, would be replaced by smaller, more fuel-efficient vehicles. I'm not even certain that a new car's MPG rating was even considered before this time, but it certainly was afterward. 

Another result of the "Gas Crisis" was the federal government's imposition of a 55 MPH maximum speed limit on all the nation's roads and highways. One of the not-necessarily intended benefits of the nationwide speed limit was the reduction of traffic injuries and fatalities, which helped put the concern for safely at the forefront of the design of new cars. 

Perhaps the most profound attitude shift of the seventies was the environmental movement which was given a great boost by nothing less than the moon missions, especially by the photographs of earth taken from outer space which showed our home as a beautiful gem in the midst of vast emptiness. Perhaps for the first time, the general public realized that although our natural resources were abundant, they are not infinite. The visits to the moon and later to Mars and Venus drove home the point that there's no remotely close, hospitable place that we can escape to if we fail to take care of our own planet home.

And speaking of the moon missions, while they were great accomplishments in their own right, it became abundantly clear that as far as space exploration is concerned, you get way more bang for the buck by sending robots into space than people. You don't have to feed them, create a cozy livable environment for them, or keep them entertained. Plus, you don't have to bring them back to earth and as everyone knows, a one-way ticket costs less than round trip. Most important, you don't have to take the unbearable risk of losing crew members, as we in the States painfully experienced with Apollo 1 and the Space Shuttle Challenger and Columbia crews.

Perhaps the final nail in the coffin of a future with bigger, faster and more comfortable means of transportation took place on the 25th of July, 2000 in Paris when a Concorde crashed just after takeoff killing everyone on board as well as four people on the ground. The result was the permanent grounding of the entire fleet of supersonic passenger jets.

Ironically, today, it typically takes us longer to get where we're going, but we're getting there safer, cheaper and with less harm to the environment, than we did fifty years ago.

While the Sci-Fi books and films of the mid-twentieth century got the means of transportation part wrong, they hit the jackpot with computers and artificial intelligence, which has made everything from fuel efficient, safe automobiles, to interplanetary space travel, to helping cure once deadly diseases to everything in between, possible.

And yes, we worry justifiably about its implications as well. 

So why do I bring all this up? 

Because about one month ago, I downloaded after considerable thought, the AI app Chat GPT. I've reached my limit with this post, so I'll have to save my report on the experience for the next post.

I'll just give you another spoiler alert, Chat GPT has left me overwhelmed, flabbergasted, blown away gob smacked and worried in an existential way, all at the same time. But is it life-changing?

Could be.

Oh and one other thing, this post has been written entirely by Chat GPT.

Seriously.

Now take me to your leader.