by Robyn Repko Waller

AI has a proclivity for exaggeration. This hallucination is integral to its success and its danger.

Much digital ink has been spilled and computational resources consumed as of late in the too rapidly advancing capacities of AI.

Large language models like GPT-4 heralded as a welcome shortcut for email, writing, and coding. Worried discussion for the implications for pedagogical assessment — how to codify and detect AI plagiarism. Open-AI image generation to rival celebrated artists and photographers. And what of the convincing deep fakes?

The convenience of using AI to innovate and make efficient our social world and health, from Tiktok to medical diagnosis and treatment. Continued calls, though, for algorithmic fairness in the use of algorithmic decision-making in finance, government, health, security, and hiring.

Newfound friends, therapists, lovers, and enemies of an artificial nature. Both triumphant and terrified exclamations and warnings of sentient, genuinely intelligent AI. Serious widespread calls for a pause in development of these AI systems. And, in reply, reports that such exclamations and calls are overblown: Doesn’t intelligence require experience? Embodiment?

These are fascinating and important matters. Still, I don’t intend to add to the much-warranted shouting. Instead, I want to draw attention to a curious, yet serious, corollary of the use of such AI systems, the emergence of artificial or machine hallucinations. By such hallucinations, folks mean the phenomenon by which AI systems, especially those driven by machine learning, generate factual inaccuracies or create new misleading or irrelevant content. I will focus on one kind of hallucination, the inherent propensity of AI to exaggerate and skew. Read more »

by Brooks Riley

‘Wenn möglich, bitte wenden.’

That was the voice of the other woman in the car, ‘If possible, please turn around.’ She was nowhere to be seen in the BMW I was riding in sometime in the early aughts, but her voice was pleasant—neutral, polite, finely modulated and real.  She was the voice of the navigation system, a precursor of the chatbot—without the chat. You couldn’t talk back to her. All she knew about you was the destination you had typed into the system.

‘Wenn möglich, bitte wenden.’

She always said this when we missed a turn, or an exit. Since we hadn’t followed her suggestion the first time, she asked us again to turn around. There were reasons not to take her advice. If we were on the autobahn, turning around might be deadly. More often, we just wanted her to find a new route to our destination.

The silence after her second directive seemed excessive—long enough for us to get the impression that she, the ‘voice’, was sulking. In reality, the silence covered the period of time the navigation system needed to calculate a new route. But to ears that were attuned to silent treatments speaking volumes, it was as if Frau GPS was mightily miffed that we hadn’t turned around.

Recent encounters with the Bing chatbot have jogged my memory of that time of relative innocence, when a bot conveyed a message, nothing more. And yet, even that simple computational interaction generated a reflex anthropomorphic response, provoked by the use of language, or in the case of the pregnant silence, the prolonged absence of it. Read more »

My wife looking at the Dolomites from the Villanderer Alm in South Tyrol the day before yesterday.

by Eric Bies

Everyone is talking about artificial intelligence. This is understandable: AI in its current capacity, which we so little understand ourselves, alternately threatens dystopia and promises utopia. We are mostly asking questions. Crucially, we are not asking so much whether the risks outweigh the rewards. That is because the relationship between the first potential and the second is laughably skewed. Most of us are already striving to thrive; whether increasingly superintelligent AI can help us do that is questionable. Whether AI can kill all humans is not.

So laymen like me potter about and conceive, no longer of Kant’s end-in-itself, but of some increasingly possible end-to-it-all. No surprise, then, when notions of the parallel and the alternate grow more and more conspicuous in our cultural moment. We long for other science-fictional universes.

Happily, then, did the news of the FDA’s approval of one company’s proprietary blend of “cultured chicken cell material” greet my wonky eyes last month.

“Too much time and money has been poured into the research and development of plant-based meat,” I thought. “It’s time we focused our attention on meat-based meat.”

When I shared this milestone with my students—most of them high school freshmen—opinions were split. Like AI, lab-grown meat was quick to take on the Janus-faced contour of promise and threat. The regulatory thumbs-up to GOOD Meat’s synthetic chicken breast was, for some students, evidence of our steady march, aided by science, into a sustainable future. For other students, it was yet another chromium-plated creepy-crawly, an omen of more bad to come. Read more »

Trying to make Sense of Red
—A Tennessee Cleanup

Its janitors are sweeping up its sins—
senators are on the floor with whisks and
fine-toothed combs. They crawl and sift,
scooping, collecting photographs
of those they’ve lynched
they cram them into
rubbish bins

—before their kids get wind

they ban   their two-faced history,
they ban   before their children come to know,
they ban   before their jittering pot lid blows

—but

maybe    they skew and hack the rules
..  ………  to save kids from what history shows,
maybe    they obfuscate for mercy’s sake,
maybe.   before their children come to see and judge,
maybe.. .they’d rather have them shot in schools

Jim Culleny, 4/8/23

by Mindy Clegg

In their oft-cited classic examination of the modern mass media, Manufacturing Consent, Edward Herman and Noam Chomsky described modern American news media thusly: “The mass media serve as a system for communicating messages and symbols to the general populace. It is their function to amuse, entertain, and inform, and to inculcate individuals with the values, beliefs, and codes of behavior that will integrate them into the institutional structures of the larger society. In a world of concentrated wealth and major conflicts of class interest, to fulfill this role requires systematic propaganda.”¹ In other words, democratic states use privately-owned media as a means of social control. Private corporations own and operate media outlets and they work with the US government because the power of the state dovetailed with their own economic interests.

The groundwork for this state of affairs emerged out of intellectual discourse in the early days of mass media. In the wake of the first world war, prominent intellectuals like Walter Lippmann and Edward Bernays suggested a set of strategies for channeling democratic impulses expanding in the United States to better align with the wishes of the ruling classes.² Such analysis was and continues to be necessary, as many are unaware of the very real pitfalls of corporate media in democratic societie. These systems are now often globalized which shape our understanding of the past and present that we must understand in hopes of changing them. But we must also wonder if the singular focus on these systems of control lead to the feelings of hopelessness that many of us feel about our institutions these days. As much as describing what dominates us feels cathartic, focusing only on the systems of control and not on resistance makes the problem seem insurmountable. I argue that we need to look for the cracks as much as describe the problem posed by corporate medi. Understanding the democratic alternatives within and outside of the mainstream production of popular culture can help us to see these cracks. Read more »

Sughra Raza. NYC, April 2023.

Digital photograph.

by David J. Lobina

The hype surrounding Large Language Models remains unbearable when it comes to the study of human cognition, no matter what I write in this Column about the issue – doesn’t everyone read my posts? I certainly do sometimes.

Indeed, this is my fourth, successive post on the topic, having already made the points that Machine/Deep Learning approaches to Artificial Intelligence cannot be smart or sentient, that such approaches are not accounts of cognition anyway, and that when put to the test, LLMs don’t actually behave like human beings at all (where? In order: here, here, and here).[i]

But, again, no matter. Some of the overall coverage on LLMs can certainly be ludicrous (a covenant so that future, sentient computer programs have their rights protected?), and even delirious (let’s treat AI chatbots as we treat people, with radical love?), and this is without considering what some tech charlatans and politicians have said about these models. More to the point here, two recent articles from some cognitive scientists offer quite the bloated view regarding what LLMs can do and contribute to the study of language, and a discussion of where these scholars have gone wrong will, hopefully, make me sleep better at night.

One Pablo Contreras Kallens and two colleagues have it that LLMs constitute an existence proof (their choice of words) that the ability to produce grammatical language can be learned from exposure to data alone, without the need to postulate language-specific processes or even representations, with clear repercussions for cognitive science.[ii]

And one Steven Piantadosi, in a wide-ranging (and widely raging) book chapter, claims that LLMs refute Chomsky’s approach to language, and in toto no less, given that LLMs are bona fide (his choice of words) theories of language; these models have developed sui generis representations of key linguistic structures and dependencies, thereby capturing the basic dynamics of human language and constituting a clear victory for statistical learning in so doing (Contreras Kallens and co. get a tip of the hat here), and in any case Chomskyan accounts of language are not precise or formal enough, cannot be integrated with other fields of cognitive science, have not been empirically tested, and moreover…(oh, piantala).[iii] Read more »

by John Allen Paulos

Despite many people’s apocalyptic response to ChatGPT, a great deal of caution and skepticism is in order. Some of it is philosophical, some of it practical and social. Let me begin with the former.

Whatever its usefulness, we naturally wonder whether CharGPT and its near relatives understand language and, more generally, whether they demonstrate real intelligence. The Chinese Room thought experiment, a classic argument put forward by philosopher John Searle in 1980, somewhat controversially maintains that the answer is No. It is a refinement of arguments of this sort that go back to Leibniz.

In his presentation of the argument (very roughly sketched here), Searle first assumes that research in artificial intelligence has, contrary to fact, already managed to design a computer program that seems to understand Chinese. Specifically, the computer responds to inputs of Chinese characters by following the program’s humongous set of detailed instructions to generate outputs of other Chinese characters. It’s assumed that the program is so good at producing appropriate responses that even Chinese speakers find it to be indistinguishable from a human Chinese speaker. In other words, the computer program passes the so-called Turing test, but does even it really understand Chinese?

The next step in Searle’s argument asks us to imagine a man completely innocent of the Chinese language in a closed room, perhaps sitting behind a large desk in it. He is supplied with an English language version of the same elaborate set of rules and protocols the computer program itself uses for correctly manipulating Chinese characters to answer questions put to it. Moreover, the man is directed to use these rules and protocols to respond to questions written in Chinese that are submitted to him through a mail slot in in the wall of the room. Someone outside the room would likely be quite impressed with the man’s responses to the questions posed to him and what seems to be the man’s understanding of Chinese. Yet all the man is doing is blindly following the same rules that govern the way the sequences of symbols in the questions should be responded to in order to yield answers.  Clearly the man could process the Chinese questions and produce answers to them without any understanding any of the Chinese writing. Finally, Searle drops the mic by maintaining that both the man and the computer itself have no knowledge or understanding of Chinese. Read more »

by Mike Bendzela

[This will be a two-part essay.]

Ischemia

When the burly, bearded young man climbs into the bed with my husband, I scooch up in my plastic chair to get a better view. On a computer screen nearby, I swear I am seeing, in grainy black-and-white, a deep-sea creature, pulsing. There is a rhythmic barking sound, like an angry dog in the distance. With lights dimmed and curtains drawn in this mere alcove of a room, the effect is most unsettling. That barking sea creature would be Don’s cardiac muscle.

It is shocking to see him out of his work boots, dungarees, suspenders, and black vest, wearing instead a wraparound kitchen curtain for a garment. He remains logy and quiet while the young man holds a transducer against his chest and sounds the depths of his old heart, inspecting valves, ventricles, and vessels for signs of blood clots. This echocardiogram is part of the protocol, even though they are pretty sure the stroke has been caused by atherosclerosis in a cerebral artery.

The irony of someone like Don being held in such a room, amidst all this high-tech equipment, is staggering. He is a traditional cabinetmaker by trade and an enthusiast of 19^th century technologies, especially plumbing systems and mechanical farm equipment. He embarked on a career as an Industrial Arts teacher in Maine in the 1970s but abandoned that gig during his student teaching days when he decided it was “mostly babysitting, not teaching.” The final break came when he discovered that one of his students could not even write his own name, and his superiors just said, “Move him along.”

In the dim quiet, while the technician probes Don’s chest, I mull over the conversation we just had with two male physicians. They had come into the room and introduced themselves as neurologists—Doctors Frick & Frack, for all I remember. Read more »

Fish in a pond in Franzensfeste, South Tyrol.

by Mary Hrovat

McCormick’s Creek State Park is one of my favorite hiking spots. The creek flows through a little canyon with a waterfall in a beautiful wooded area. I’ve been visiting the park for more than 40 years. It’s a constant in my life, whether the waterfall is roaring in flood or slowed to a trickle during a dry spell or, once in a great while, frozen solid. 

Late in the evening of Friday, March 31, an EF3 tornado struck the campground in the park. It caused considerable damage, and two people were killed. After the tornado left the park, it seriously damaged several homes in a rural area just outside Bloomington. It was part of an outbreak of 22 tornadoes throughout the state. A tornado that hit Sullivan, Indiana, destroyed or damaged homes and killed three people. 

It was tough to see spring beginning with such serious damage and loss of life in a beloved spot. It was also sobering to see photographs of the destruction in Sullivan. It seems that I’ve seen many such images from places to the south and southwest of us this winter, and in fact 2023 has been an unusually active year for tornadoes in the U.S. so far. There have already been more tornado fatalities in 2023 than in all of 2022 nationwide.  Read more »

Prophets on the Nairobi Expressway

by Rafiq Kathwari

“Please take the next flight to Nairobi,”
my niece said, her voice cracking over

WhatsApp. “Mom is in ICU. Lemme know
what time your flight lands. I’ll send the car.”

Early February morning on the Upper West Side,
I wore a parka, pashmina scarf, cap, gloves, rode

the A-Train to JFK, boarded Kenya Airways,
and 12 hours later

even before we landed at NBO, I peeled off my
layers anticipating equatorial warmth, the sun

at its peak, mid-afternoon. I waved at a tall, lean
man holding up RAFIKI scrawled on cardboard.

“Welcome,” he said.
“What’s your name?” I asked.

“Moses,” he said as we flew on the Expressway,
built by the Chinese.

“Oh,” I said. “My middle name is Mohammed.
Let’s look for Jesus and resurrect my sister.”

by Ali Minai

How intelligent is ChatGPT? That question has loomed large ever since OpenAI released the chatbot late in 2022. The simple answer to the question is, “No, it is not intelligent at all.” That is the answer that AI researchers, philosophers, linguists, and cognitive scientists have more or less reached a consensus on. Even ChatGPT itself will admit this if it is in the mood. However, it’s worth digging a little deeper into this issue – to look at the sense in which ChatGPT and other large language models (LLMs) are or are not intelligent, where they might lead, and what risks they might pose regardless of whether they are intelligent. In this article, I make two arguments. First, that, while LLMs like ChatGPT are not anywhere near achieving true intelligence, they represent significant progress towards it. And second, that, in spite of – or perhaps even because of – their lack of intelligence, LLMs pose very serious immediate and long-term risks. To understand these points, one must begin by considering what LLMs do, and how they do it.

Not Your Typical Autocomplete

As their name implies, LLMs focus on language. In particular, given a prompt – or context – an LLM tries to generate a sequence of sensible continuations. For example, given the context “It was the best of times; it was the”, the system might generate “worst” as the next word, and then, with the updated context “It was the best of times; it was the worst”, it might generate the next word, “of” and then “times”. However, it could, in principle, have generated some other plausible continuation, such as “It was the best of times; it was the beginning of spring in the valley” (though, in practice, it rarely does because it knows Dickens too well). This process of generating continuation words one by one and feeding them back to generate the next one is called autoregression, and today’s LLMs are autoregressive text generators (in fact, LLMs generate partial words called tokens which are then combined into words, but that need not concern us here.) To us – familiar with the nature and complexity of language – this seems to be an absurdly unnatural way to produce linguistic expression. After all, real human discourse is messy and complicated, with ambiguous references, nested clauses, varied syntax, double meanings, etc. No human would concede that they generate their utterances sequentially, one word at a time. Read more »

Chuang Tzu’s Butterfly

“Once upon a time, I, Zhuangzi (Chuang Tzu), dreamt I was a butterfly, fluttering hither and thither, to all intents and purposes a butterfly. I was conscious only of my happiness as a butterfly, unaware that I was Zhuangzi. Soon I awakened, and there I was, veritably myself again. Now I do not know whether I was then a man dreaming I was a butterfly, or whether I am now a butterfly, dreaming I am a man. Between a man and a butterfly there is necessarily a distinction. The transition is called the transformation of material things.” —Chinese poet/philosopher, 4^th century BC

Chuang Tzu’s Butterfly

the other night when I was sleeping
gone so far the moonlight leaping
through my window, past the curtain,
instantly I knew for certain
that I was a butterfly

I went flitting flower to flower,
I grew freer by the hour,
no concern for job or romance,
through the night I just
danced and I danced

but when the morning light was breaking,
the sun, the sun, the moon forsaken,
I got up threw back the covers,
instantly I was another
knew I was a man again

between that butterfly and me
I must make some kind of line,
can’t have a common destiny

between me and this lungful of air that I breathe
I must make some kind of line
something solid my reason can squeeze

Jim Culleny
(Written as a song in 1975)

by Thomas R. Wells

Environmentalists are always complaining that governments are obsessed with GDP and economic growth, and that this is a bad thing because economic growth is bad for the environment. They are partly right but mostly wrong. First, while governments talk about GDP a lot, that does not mean that they actually prioritise economic growth. Second, properly understood  economic growth is a great and wonderful thing that we should want more of.

Governments around the world – of every ideology – are in favour of economic growth all else being equal. Economic growth increases the wealth of a population and hence improves their options and those of the government that rules them. This is extremely politically convenient as it allows governments to serve all their various constituencies without having to make hard choices between them, and so keep them happy enough that they get to stay in power. Honest politicians can provide more public services to those who demand them, while keeping the tax rate the same. Corrupt politicians can get away with funneling money to themselves and their cronies without risking revolution. More money means fewer and easier political problems.

However, just because someone values a certain outcome, does not mean that they value it enough to take the necessary painful steps to achieve it. (Or else everyone would get A’s in all their exams and keep the waist size they had in high-school.) It turns out that the policies governments need to implement (or stop implementing) in order for their societies to get richer are often more politically costly than they are worth. Take for example governments’ responsibility for the housing crisis across the rich world, in which the price of housing rises faster than incomes. Read more »