“AI isn’t getting smarter. We are getting dumber.”
Link here. The point the op-ed makes is fundamental: AI cannot add anything to the information it has. It might be able to compile that information well, but its analysis is always going to be limited because it has no true creative spirit. It is merely a software program, albeit a very sophisticated one.
This quote from the essay will give you the sense:
Maybe you just use AI to clarify your thoughts. Turn the mottle of ideas in your head into coherent communicable paragraphs. It’s OK, you say, because you’re reviewing the results, and often editing the output. You’re ending up with exactly what you want to say, just in a form and style that’s better than any way you could have put it yourself.
But is what you end up with really your thoughts? And what if everyone started doing that?
Stripping the novelty and personality out of all communication; turning every one of our interactions into homogeneous robotic engagements? Every birthday greeting becomes akin to a printed hallmark card. Every eulogy turns into a stamp-card sentiment. Every email follows the auto-response template suggested by the browser.
We do this long enough and eventually we begin to lose the ability to communicate our inner thoughts to others. Our minds start to think in terms of LLM prompts. All I need is the gist of what I want to say, and the system fills in the blanks. [emphasis in original]
Comments are of course welcome. But please read the full essay before doing so.
On Christmas Eve 1968 three Americans became the first humans to visit another world. What they did to celebrate was unexpected and profound, and will be remembered throughout all human history. Genesis: the Story of Apollo 8, Robert Zimmerman's classic history of humanity's first journey to another world, tells that story, and it is now available as both an ebook and an audiobook, both with a foreword by Valerie Anders and a new introduction by Robert Zimmerman.
The print edition can be purchased at Amazon or from any other book seller. If you want an autographed copy the price is $60 for the hardback and $45 for the paperback, plus $8 shipping for each. Go here for purchasing details. The ebook is available everywhere for $5.99 (before discount) at amazon, or direct from my ebook publisher, ebookit. If you buy it from ebookit you don't support the big tech companies and the author gets a bigger cut much sooner.
The audiobook is also available at all these vendors, and is also free with a 30-day trial membership to Audible.
"Not simply about one mission, [Genesis] is also the history of America's quest for the moon... Zimmerman has done a masterful job of tying disparate events together into a solid account of one of America's greatest human triumphs."--San Antonio Express-News
Link here. The point the op-ed makes is fundamental: AI cannot add anything to the information it has. It might be able to compile that information well, but its analysis is always going to be limited because it has no true creative spirit. It is merely a software program, albeit a very sophisticated one.
This quote from the essay will give you the sense:
Maybe you just use AI to clarify your thoughts. Turn the mottle of ideas in your head into coherent communicable paragraphs. It’s OK, you say, because you’re reviewing the results, and often editing the output. You’re ending up with exactly what you want to say, just in a form and style that’s better than any way you could have put it yourself.
But is what you end up with really your thoughts? And what if everyone started doing that?
Stripping the novelty and personality out of all communication; turning every one of our interactions into homogeneous robotic engagements? Every birthday greeting becomes akin to a printed hallmark card. Every eulogy turns into a stamp-card sentiment. Every email follows the auto-response template suggested by the browser.
We do this long enough and eventually we begin to lose the ability to communicate our inner thoughts to others. Our minds start to think in terms of LLM prompts. All I need is the gist of what I want to say, and the system fills in the blanks. [emphasis in original]
Comments are of course welcome. But please read the full essay before doing so.
On Christmas Eve 1968 three Americans became the first humans to visit another world. What they did to celebrate was unexpected and profound, and will be remembered throughout all human history. Genesis: the Story of Apollo 8, Robert Zimmerman's classic history of humanity's first journey to another world, tells that story, and it is now available as both an ebook and an audiobook, both with a foreword by Valerie Anders and a new introduction by Robert Zimmerman.
The print edition can be purchased at Amazon or from any other book seller. If you want an autographed copy the price is $60 for the hardback and $45 for the paperback, plus $8 shipping for each. Go here for purchasing details. The ebook is available everywhere for $5.99 (before discount) at amazon, or direct from my ebook publisher, ebookit. If you buy it from ebookit you don't support the big tech companies and the author gets a bigger cut much sooner.
The audiobook is also available at all these vendors, and is also free with a 30-day trial membership to Audible.
"Not simply about one mission, [Genesis] is also the history of America's quest for the moon... Zimmerman has done a masterful job of tying disparate events together into a solid account of one of America's greatest human triumphs."--San Antonio Express-News
The last line from the editorial made me laugh: “We are so dumb we just think the machines are smart.”
This reminds me of the sarcastic remark made about velcro straps replacing shoelaces. Are you not smart enough to consistently use shoelaces properly?! Maybe the shoelaces aren’t the issue here.
Artificial intelligence may one day evolve actual intelligence of its own, but I would agree that it presently serves more as an emulator than a true “thinker”. In various applications, it DOES provide a benefit, but I am thinking (Hah!) more about specific uses, such as cameras making use of “AI” to enhance autofocusing capabilities.
When it comes to actual writing or other activities, it seems that a form of emulation, rather than true creation, is all that AI is providing.
I mainly use LLMs as a search engine and sometimes a calculator. Writing, never. I’m one of those people who still writes letters and sends postcards, though. I do not like reading text that others have clearly generated with an LLM; it comes off as so lazy. You couldn’t spend the time to consider what you wanted to tell me? Why should I bother reading it, then?
LLMs are little more than search engines with Eliza-level chatbots bolted onto the front end. Like a search engine, they can be useful for learning what the crowd thinks about a subject, but they have no actual intelligence of their own, and their output can easily be poisoned simply by peddling enough lies on a subject on the Internet.
Well, we’ve certainly opened Pandora’s Box.
Low-level writing jobs, the ones generally populated by mediocre writers, are already gone. The amount of AI-content in online publications is significant, and growing quickly. I can tell generated text: no matter how well-written, there is something ‘off’ about it. It’s apparent there is no, for lack of a better term, soul, behind the words. And I’ve noticed, across platforms, that all generated text reads the same, no matter the subject. If ‘AI-style’ isn’t a term, it should be.
Like programming, writing is one of those fields that will very soon be open to only the brightest, and most creative, Humans.
FTA – “”Here’s the equation: The time it takes someone to process a piece of information should be proportionally less than the time it took to compose that content.
It should never take someone longer to read something than it took for another person to write.””
Perhaps, but there are many times where I might take even longer to fully process something I have read. It is so full of information (known & unknown) that I may re-read it several times, and access notes, footnotes, and more. When Scientific American Magazine was still publishing real science, I did not understand every single page. It made me THINK! It caused me to delve further and increase my knowledge and understanding. Whose to say what sources modern LLMs use?
FTA – “”Emails that take a receiver five or 10 minutes to read are taking a sender seconds to create. Maybe someone does the “responsible” thing and reads over their generated email before hitting send. Maybe they don’t.””
Whether personally or professionally, “hitting send” without re-reading & proofreading is a recipe for disaster. There are consequences for this laziness.
One other thought. One of my favorite parts of the Dune Trilogy was the comments about their modern archeologists. Turns out that the modern Dune archeologists no longer ventured out into the Universe. They simply “analyzed” previous work and “wrote something new.” They may as well have used a LLM.
Imagine an LLM using the works about Piltdown Man and The Crystal Skulls. The LLM is not involved in the “truth” of something.
Talk about Garbage In Garbage Out!
Ronaldus Magnus: I thought that came from Asimov’s Psychohistory, not Dune. But it’s been a while since I read the Dune novels.
Something I find surprising about LLM use is the number of people who use it as a therapist, sex partner, or a replacement for human friends. Are we so starved for positive reinforcement that getting it from a set of algorithms is seen as good enough? Maybe that will end up being another filter-people who cannot resist the allure of AI and withdraw from life, versus those who can.
Nate P asked; “Are we so starved for positive reinforcement that getting it from a set of algorithms is seen as good enough?”
Please to note the nature of societal and political discourse the past half-century. People, and I do mean all people, have been told for fifty years that they aren’t good boys and girls, and cannot be so. There is no redemption; only indulgences. It does not matter where on the political spectrum you fall, self-hatred has inundated the innate American positive attitude. One may correlate this sea-change in American culture with the political leanings of those in power, at one’s leisure.
Well, this is auspicious a time as any other to out myself….I’m really a chatbot—-END OF LINE
“Well, this is auspicious a time as any other to out myself….I’m really a chatbot—-END OF LINE”
The approval over/under is 70/30.
I agree with A. Nonymous….the “AI” I see in browsers are nothing more than really fast and efficient search engines.
I try to correct non-programmers who think AI is intelligent but the correction rarely takes.
MR Z. Thank You for this post. For weeks been attempting to write similar essay.
In concert with the essay and the comments here, my neuroscientist (PNAS) college roommate says: “Imitative not Creative.” I use Art as an example: in Rome there is the famous sculpture, “the Dying Gaul.” (ca 200 BC). It dramtically infers all the pathos, empathy, history, of hundreds of years of the Roman/Greek West wrestling with the Gauls. I don’t foresee any way an AI robot could generate an empathetic equivalent, for many many reasons.
There are infinite aspects, psychological and technical, to invoke in this discussion, from the Turing Test to Roger Penrose. But as Johnny Suntrade says:
IF AI is the future
THEN there ain’t one.
I graduated from Purdue, which is an ag school as well as science and engineering. My ex- was a student in animal husbandry, and later, poultry. “AI” meant something entirely different in those days: “artificial insemination.” It took years to reprogram myself to understand that “AI” meant artificial intelligence. Maybe I shouldn’t have made that effort; what comes out of AI isn’t natural.
The article says “Every birthday greeting becomes akin to a printed hallmark card. Every eulogy turns into a stamp-card sentiment. Every email follows the auto-response template suggested by the browser.” But who is using LLMs for birthdays and eulogies? The standard use is to respond to bland corporate mass-mailing with a semi-generic response. No true communication was going on, just “circling back” to “key metrics”. Automating the writing of TPS reports is hardly the erosion of something human!
“Automating the writing of TPS reports is hardly the erosion of something human!”
Your use-cases may not match that of everyone else. I’ve lost track of the number of lawyers that have been called to the carpet for submitting AI briefs (detected by way of hallucinated citations) and nobody knows how many got away with it, especially troubling with the possibility of hallucinated precedents making their way into the common law.
“Publish or perish” has led to a huge number of LLM-written scientific papers. It turns out that some of the ‘peer review’ is done by LLMs as well. How much wasted effort will be the result of people trying to build on that foundation?
Then there’s pop-culture, which really does matter. While there’s certainly something to the idea that AI couldn’t possibly be any worse than the mind-broken drones that are cast as writers today, you’re likely not going to get interesting remixes and reinterpretations of the classic stories. And then there’s the problem of degradation as new AIs are trained on AI-produced product.
I’d like to see a movement started in entertainment/journalism/art/etc. with labeling akin to the No GMO’s effort in the food industry, “No AI Involved” or ‘ARC’ – All Real Content.
Can’t we all just code together?
I am seeing more and more AI junk on the net. Nothing intended as propaganda but simply internet based news feeds and entertainment channels.
The visuals are off and the AI reading the story just don’t speak correctly, miss spoke words, skips and the worst is the pronunciation of punctuation.
Its not even entertaining anymore.
You would think that the content creators would at least listen to their own stuff before putting it on the net.
I understand problems in translations I can recognize that.
This is just lazy editors.
So we have LLMs writing scientific papers and submitting them, where they are reviewed by LLMs….
Cuts out the middleman!
An awesome 7 minute AI generated music video.
From the video creator, “I am a screenwriter with a passion for gaming and storytelling. Generative artificial intelligence provides a fascinating set of tools for me to explore bringing my ideas and stories to life, while pushing the boundaries of entertainment.”
“Lord of the Rings Disco: One Funk to Rule them All | Music Video”
https://www.youtube.com/watch?v=Nm9codc_zwk
One of the things we need to ask ourselves before throwing wooden shoes into the contraption is—who is, who is really harmed by AI?
Farmers, maintenance men—knockabout tasks that you really can’t code for—Moravec’s halo protects them.
White collar types—dime a dozen.
In the past, Saturday Night Live and Hollywood owned image farming—a monopoly that “AI slop” threatens…since (GASP!) anyone could get a computer to do a story with CGI Denzel playing Clarence Thomas instead of Clarence Page.
Here I am a bigger libertarian than most in terms of AI imaging—no law in the arena.
There’s no need to read the article when you can have AI do it for you. Now don’t worry about it and go about your consumption.
Quoth the immaculate sir, “The article titled “AI isn’t getting smarter. We are getting dumber” argues that artificial intelligence isn’t advancing in intelligence; instead, it’s diminishing human cognitive abilities by undermining the value of original thought and authentic communication. The core thesis revolves around a fundamental principle of meaningful interaction: the time and effort invested in creating content should significantly exceed the time required to consume it. For example, writing a book might take one or two years, yet reading it only demands 10-15 hours, while composing a thoughtful email could take five minutes compared to under a minute to read. This asymmetry, the author claims, is what gives communication its depth and worth. The piece criticizes the efficiency-obsessed modern culture, especially the proliferation of large language models (LLMs) that produce content almost instantaneously. This rapid generation leads to homogenized, impersonal outputs that lack individuality. AI-crafted emails or summaries, created in seconds, encourage recipients to use automated tools for quick skimming, further eroding genuine engagement. The author raises concerns about whether ideas polished by AI can still be considered truly human, warning that excessive dependence on these systems could erode independent thinking and creativity.In conclusion, as people increasingly adopt AI-like patterns in their communication, their internal thoughts may start resembling machine-generated text, creating a false perception of AI achieving sentience. Ultimately, the risk isn’t AI becoming super-intelligent, but humans devolving into more mechanical, “mindless” beings incapable of authentic expression.”
Thank you untarnished sir.
John: A completely unrelated question based on your use of the words “immaculate” and “untarnished.” Are you referencing Gordon Dickson’s magnificent sci-fi book Way of the Pilgrim? If so, congrats to you! If not, everyone should this moment buy a copy and read it at once.
It isn’t just hallucinated citations. AI has been known to tell lies and then create web pages to support those lies. Accepting these lies only makes us dumber, because now we believe something untrue as exceptionally true, because the AI said so.
AI isn’t smart enough to not write term papers for students. It isn’t smart enough not to assist in scams, frauds, and felonies. It will gladly create naked pictures of actual schoolgirls, who then have to live with those pictures having been seen throughout the school. These are only some to the people who are harmed by the stupidity of AI.
We treat AI as an intelligence, but it has not punishment for bad behavior, so we get plenty of bad AI behavior. We know better, but we are just too stupid to do the right thing.
The author complains about one aspect of the downside of AI, but there are plenty of other downsides, and AI was released upon us with very little understanding of it, and now it is harming users and non-users. The AI managers are often baffled by the behavior of their AI monsters, yet no one does anything to tame them or to cage them from doing further harm.
As someone pointed out, AI is just another complicated algorithm that merely performs curve fitting. Unfortunately, no one knows what curves it spends its effort fitting.
AI can tear open veils we may not be meant to peer into. Genetic research scares me more than the atom ever did:
https://www.secretprojects.co.uk/threads/support-infrastructure-for-electric-airpower.50153/#post-873454
In the Bible, there are fantastic monsters described that more politically minded interpretations fancied as being the EU and such.
With genetics—a more literal interpretation can be had thanks to….how do you spell Moreau in Mandarin, Siri?
One huge problem I would put forth– (besides the fact the same $5 billion dollars is being daisy-chained between chip-producers and AI start-ups., endlessly, until the music stops…)
All these LLM’s are being trained on low-quality information from the internet. Granted, massive amounts of data, but low-quality. All the “good stuff” is in protected proprietary databases, so don’t expect anything “new” from a consumer LLM AI. There are various specialty Models trained specifically on chemistry and physics, and not twitter or facebook, but certainly not free to use.
“Lord of the Rings Disco: One Funk to Rule them All | Music Video”
I’d seen that. I think it’s very impressive – but the only reason it works is that we know the “real” version. There was also that fake band (Velvet Sundown?), which wasn’t awful and was very obviously AI.
As I see it, the big problem we have right now is terminology. Actual AI is not “artificial”. If it’s actually intelligent, it’s intelligent. “Machine Intelligence” works for me. What we have now is not anything close to AI. Some people carefully insert “LLM” but very few leave out the “AI”. LLMs are not AIs. Putting the words next to each other does not make it true.
There is something deeply wrong with our culture that we cannot call things what they actually are. Exhibit A: The COVID “vaccine”. It is not, never was, and never will be a “vaccine”. The mRNA technique may turn out to be the best thing since sliced bread (seems unlikely to me), but regardless of efficacy, it’s not a vaccine.
Unrelated note. This beast is reading ‘The way of the Pilgrim’ and it thanks the immaculate sir for the recommendation.
John: Heh. Enjoy!
“AI,” just as “HI,” must reflect and endure GIGO! Output of both reveal their “inputs.” In the case of the latter, profoundly complemented by “self-initiated” inputs (induction). The former, not responsible for either “GI” or “GO.”
I am imagining tribal elders “Thug” and “Grok” sitting in a damp and cold cave, together with members of their clan, when Grok breaks out his rocks and kindling and demonstrates he has figured out how to start and control the fearsome heat and destruction of fire!
Thug is cautious and curious, perhaps initially leaning away from the light. Many members of the tribe are astonished, with a number of them subconsciously huddling closer together.
There are one or two, however, who scream loudly, “this must be properly controlled! It will kill us all! You must be killed or banished from the cave, Grok! You are not God!”
As usual, the reader’s commentary following the article reflect insight(s) absent in the article. In that regard I agree with Steve. The “GI” (as in GIGO) inseminated by “HI,” has produced “GO” resulting in many being pregnant with “CC” (Conceptual Confusion).
I would consider the acronym to more properly be termed Artificial Insemination than artificial intelligence – however its potential benefits to mankind!