#industrial automation programming languages

[xiaq]

Let’s talk about AI (the good, the bad, the ugly)

Hi, I work in tech and I’m also an author and I feel we need to have a chat. Because I’m seeing a lot of misinformation/conflation happening.

There are many different kinds of AI, and disagreement about what should even be termed “AI” in the industry (it’s the hot new thing, so companies are rebranding all sorts of products they offer as “AI” which is muddying the waters re what is Artificial Intelligence and what’s just a sparkling workflow). That being said, let’s go over some key terminology/how these iterations differ (at least in my part of the data management AI world).

Narrow AI: This is a limited, passive, task-specific kind of AI. It’s programmed to respond within certain constraints. Narrow AI is behind the chat bot you talk to when you want to make an Amazon return, or voice assistants you speak to when trying to make an insurance claim. They don’t learn; if you ask them a question outside their purview, they will not provide an answer. Narrow AI can be useful for efficiency and saving costs (it can also be hella annoying if you’re stuck speaking to one that can’t help you with your problem and you don’t know the magic set of words to get to an actual person who can).

Agentic AI: This is a more advanced system that attempts to mimic human decision-making within a specific context using LLMs. This is a model that can “learn” and adapt. There are many subsets of Machine Learning within Agentic AI like Supervised learning—where a system can be trained to identify something by certain characteristics (used for cancer detection!) or Unsupervised Learning which is a matching/pattern recognition exercise (like identifying new disease subgroups!). Agentic AI can also help more generally with looking at patient data within the context of their medical history/the most up-to-date medical best practices, and provide insights. Aside from use in the medical industry, Agentic AI might be used within larger corporations for supply chain management—monitoring and automating interactions between suppliers, vendors, freight companies, etc. to make sure the correct number of products are ordered and shipped at the correct time to the correct locations, even if those numbers fluctuate. It can be used for companies who want to enable self-service for their business units to query data and create new data sets based on those queries using natural language (“show me all customers who purchased x product within this time frame in this geographic location,” “now create a new data set with this information”). Like Narrow AI, Agentic AI can improve efficiency, and is helpful in contexts when the breadth of data/moving parts involved is so substantial a human may not be equipped to manage it alone. However, it still needs to be implemented responsibly (more on that after Generative AI).

Predictive AI: This uses pattern recognition/machine learning/statistics/algorithms to predict outcomes. Predictive AI taps into an amount of data that was previously not possible for human teams to manage (much like Agentic AI). Predictive AI can anticipate stock market changes, extreme weather, mechanical issues, supply chain impacts, healthcare outcomes, crime surges, and more, saving time, money, and potentially even preventing major problems like vehicle recalls and death due to natural disasters. However, predictive AI is limited by the data it’s trained on, which has resulted in algorithmic biases (like when used in law enforcement/policing contexts, or healthcare contexts). So, as is true for any of the AI models I’ve mentioned so far, while it can be a positive tool, implementors should be cognizant of the fallible, human, foundation it’s built upon and try to mitigate bias.

Generative AI: This is what most people think of when they think of AI, thanks to chatbots like ChatGPT. Generative AI is a regurgitative leech. It creates “new” content based on the massive amounts of data it has been trained on. Nothing ChatGPT creates is actually new, though. It’s not thinking for itself. When you ask it to write a story or create a picture, it’s using an amalgamation of the writing and art it has copied from real creators without credit. There is very little useful about Generative AI like ChatGPT. Especially when you consider the environmental ramifications of using it. Generative AI used within a social context (as a therapist/friend/romantic stand-in) is dangerous. And if used as an authoritative search engine (which is disturbingly prevalent, now) it’s equally problematic due to common issues like hallucinations, the spread of fabricated news stories/outlets, dangerous deepfakes, extremist bias, and more. And if you’re thinking, well I just use it to help with outlining papers/re-writing emails/condensing notes, there are already studies that raise concerns about generative AI weakening critical thinking skills. I cannot tell you how relieved I am that I left my job as a professor the year before ChatGPT came out.

Now, to be fair, a lot of Agentic AI depends on LLMs, as I mentioned, which makes it prone to encountering the same sorts of issues with hallucinations/bias as Generative AI chatbots. But in my experience, companies/hospitals/government entities are aware of this and, when implementing responsibly, use Agentic AI as a tool that requires supervision and adjustment, not some holy, infallible, authority (like many public-facing chat bot users). I also think the potential benefits of Agentic AI currently outweigh my concerns about its use of generative AI (though the environmental impacts are still worrisome). So perhaps even further nuance is needed between Generative AI used within Agentic contexts and public-facing Generative AI used purely for entertainment/ “education” in chatbots.

Which only emphasizes my point that lumping all AI together is not beneficial. There’s nuance! Anyway.

TL;DR my personal thoughts on AI:

Narrow AI—Can be Good when implemented appropriately!

Agentic AI—Can be Good when implemented appropriately!

Predictive AI—Can be Good when implemented appropriately!

Generative AI (public-facing)—Kill it with fire.

[ohnoitstbskyen]

I had a few requests to repost this as its own rebloggable post, so here you go. The original person I was responding to deactivated their reblogs, presumably because they got a lot of unpleasantness in the notes. This is practising healthy online boundaries, and I'd appreciate it if people don't give them any sh** for it.

---------------------------------

Hm, there are at least three major problems to consider, I think. This is going to be a bit long, so bear with me, but I promise that I will not at any point argue in favor of copyright, argue on the grounds of human ability, and I will not at any point appeal to concepts of "soul" or "spirit."

1: Environmental impact

It's fairly well documented at this point that the sheer power requirements to run AI data centers and computation have caused the major tech companies betting the farm on this technology to invest in a huge expansion of data centers, which in turn both require huge amounts of additional power (which tends to be drawn from fossil fuel sources) and huge amounts of water for cooling.

For reasons unrelated to AI as a technology as such, but related to the capitalist mode of production which produces "AI" as a product and service, the burdens and costs of creating this new infrastructure falls disproportionately on the poor, the marginalized and upon the global south, as it always does. It is of course to be noted that these tech companies are likely to also be using generative AI as an excuse to backslide on their environmental commitments and massively expand their infrastructure with government and venture capital money, but generative AI is a not-insignificant part of the problem.

2: Economic impact

The primary stated purpose of generative AI as a business is to replace workers. It is a form of automation, and generative AI specifically targets jobs requiring language and visual media skills - whether that be translation, copywriting, creative writing, coding, drawing, painting or graphic design, or any of a thousand other related skills.

The express purpose of adopting the technology widely is to replace hundreds of thousands of workers, who upon losing their jobs will be thrust into precarity, and the industries affected will experience an enormous downwards pressure on wages and an enormous negative impact on processes of unionization and collective bargaining.

One might argue that this is a problem of capitalism more than a problem of generative AI as a technology, which, okay, sure, but capitalism is the system we live under and the value and ethical status of any technology is always evaluated in the context of the system that deploys it. If we lived under Luxury Gay Space Communism, I am sure I would feel different about generative AI; but we don't, so I judge it accordingly. Under the capitalist economic mode of production, generative AI is a fundamental threat to the economic and social well-being of hundreds and hundreds of thousands of workers in hundreds of fields of labor.

3: Freedom of expression

Generative AI as a tool imposes harsh restrictions on freedom of expression, and in a worst-case scenario (i.e. the exact scenario that the developers of these AI tools are trying to make happen), will put huge swaths of artistic expression under the control and economic exploitation of enormous corporations unaccountable to any democratic power.

No generative AI tool can generate any output except that which is allowed for

by its programming as defined by its owners and creators

by its dataset, which must be built and maintained by an organized group

With every single major generative model currently firmly under the control of corporate or national interests, those groups have the ability to exert profound control over which outputs are or are not possible from the most powerful and most capable AI models. The Chinese DeepSeek model, for example, will refuse to answer questions about Tianamen Square, and similar censorship can and will be implemented by other state actors when it suits their purposes.

No generative model can output a picture of a motorcycle unless the database for the model is trained on pictures of motorcycles, and even if it IS trained on pictures of motorcycles, it can and will only ever output motorcycles identical to whatever necessarily limited set of motorcycles are represented in its data. And if a corporation (or government) decides that a model is not allowed to output images of motorcycles, they can implement profound censorship of the concept of motorcycles in their models.

Replace "motorcycle" with "sexual education material," "political literature" or "information about or depictions of queer people or minorities" and you start to see the problem.

One might argue that "users would simply circumvent that restriction," but to say so is to miss the point entirely: savvy power users with an agenda would circumvent the system. Your grandma using the system casually would not, especially if (or when) circumventing the system is made illegal.

Similarly, one might argue that a band of dedicated, democratically minded individuals could simply band together to train and create their own independent models, free of all censorship - but again, the ability to access the infrastructure necessary to build such a project would be contingent on the assent of either corporations or national governments, and both of those groups would inevitably see a completely unbound, democratically governed AI infrastructure as a direct competitor or a political threat, and act accordingly.

Generative AI, logistically and structurally, is a tool of expression which is privileges power. Whoever holds power in a given society has undue ability to influence what it is possible to express via generative AI and who has access to the ability to express it.

Generative AI has been magnanimously made available for "free" thus far, by speculative corporations backed by oceans of venture capital, but they are all expected to turn a profit at some point, and once those screws come down, not only will they silo and closed-source their technology, they will aggressively pursue hostile action towards competitors, and limit access to their Revolutionary™ technology to whoever is most able to either pay or coerce access through force.

In short, whoever has the most money or political power will have access to the greatest degree of freedom of expression from generative AI.

3b: Freedom of expression on the purely aesthetic level

This is a less important objection than point 3 above, but it also needs to be noted that generative AI as a technology is fundamentally based upon and limited by probability.

That is to say, when you prompt a generative AI model to generate a given output, the fundamental nature of the math it uses is a probabilistic attempt to approximate an acceptable answer.

In oversimplified terms, if you ask it for a picture of a motorcycle, it will attempt to output a semi-randomized mixture of all its data which is tagged with "motorcycle," weighted against training data and a history of user feedback, in order to probabilistically arrive at whatever output it calculates the end user is most likely to accept and validate as a "correct" output.

This has some consequences, specifically that generative AI is fundamentally bound as a technology to always and forever regress to the mean. It will always and forever default to and privilege lowest common denominators.

Thus, one of the things generative AI tends to have a lot of trouble with is ugliness. It struggles to generate images which do not conform to dominant standards of beauty and desirability. And I mean this not just in terms of human beauty and desirability, but beauty and desirability across all forms of expression. Generative AI severely struggles to output an ugly landscape image, for example. It struggles to output pictures of kittens that are not cute, it struggles to output pictures of Ferraris that don't look gorgeous or swords that don't look cool. It also struggles to output images of people along the same lines.

The first part of the problem is that the vast, overwhelming amount of input of images bias towards the aesthetically pleasing (because most images that get preserved and uploaded of anything bias towards the aesthetically pleasing). The second part is that, the vast, overwhelming amount of desired and validated outputs also bias towards the aesthetically pleasing, because that is what most users most of the time will inevitably and statistically want and prefer.

Because the vast, overwhelming majority of the model's feedback from users will validate and confirm the correctness of outputs that bias beauty and appeal, models will always be statistically required to generate outputs towards that standard.

The result is that generative AI is, technologically and irrevocably, bound to bias towards and reproduce the lowest common denominator, and it uncritically inherits every single cultural bias of the dominant culture that produces and uses it. Usually this means biases in favor of beauty and appeal, and biases against whatever is considered ugly or undesirable in a culture.

Even if you are the sort of profoundly shallow and unimaginative person who believes that "nobody would ever WANT images that aren't beautiful and appealing," you have to concede that any tool which does not allow you full and equal freedom to depict the ugly and the unconventional is imposing a severe restriction of the freedom of expression of its users.

In conclusion: generative AI is defined by mounting and catastrophic social, economic and environmental costs, and fundamentally and structurally biased in favor of power and capital. Even in the very best-case possible scenario in which all economic and political problems with the technology are solved, its fundamental nature is to regress to the mean and privilege the lowest common denominator, imposing inherent restrictions on freedom of expression. None of these problems are shared by a pencil, or by a typewriter.

[eubomubo]

| Midnight Noise |

Chapter 5

Authors note: Masterlist is out!

P.S this chapter was inspired by Evergreen by Richy Mitch & The Coal Miners <3

WARNINGS ➺ 18+ minors dni, explicit language, mental health themes (bipolar disorder, breakdown mentioned), substance use (smoking, alcohol), toxic behavior, sleep deprivation, mutual antagonism, enemies to lovers, reader is intentionally unlikeable at first, service industry trauma, emotional dysregulation, description of mental health crisis (past), blacklisting/career trauma, Eventual smut.

DISCLAIMER ➺ This story is fiction, and it does not reflect real life in any way.

WORDS ➺ 2.5k

———————————————————————————-

The L rattled through the tunnel, each lurch making the stain on her dress release fresh waves of tannin smell. She sat in the corner of the car, arms crossed over the stain, watching Chicago blur past in the dark windows.

Her phone buzzed.

Tony: Don't bother coming back. Check will be mailed.

She stared at it until the screen went dark. Then lit it up and read it again.

That was it. Three years of closing shifts, dealing with handsy customers, pretending to laugh at Tony's racist jokes, ended in a text before she'd even made it home.

The couple across from her kept sneaking glances. She must look deranged, dress wet, hair sticking to her face, clutching her phone like a vice.

Another buzz.

Tony: Very disappointed.

Very disappointed. Like she was a kid who broke curfew.

She typed out a response and deleted it. What was the point? Tony had already made up his mind the second Jake told him what happened. Probably called him before she'd even left The Bear.

The automated voice announced her stop. She stood, legs shaky, the dress peeling away from the plastic seat with a wet sound that made the couple look back at her again.

The hallway was empty. Thank fuck. She didn't have the energy to take the piss stairs.

Inside her apartment, she peeled off the dress, leaving it in a pile by the door. The shower was scalding, she sat under it with her knees drawn to her chest until her fingers pruned.

She'd been fired over text. While covered in wine. After being humiliated in front of an entire restaurant.

She pressed her forehead against the cool tile.

Tomorrow she'd have to look for a new job. Tonight she just needed to not think about Carmy telling her to get the fuck out.

———————————————————————-

She had finally landed an interview at Applebees.

She was desperate enough to feel elated when they called.

The manager's office smelled like Febreeze. She sat in a plastic chair that squeaked every time she shifted.

"So." He squinted at her resume. Cole. His name tag was flipped backward.

"Says here you were a... somm-el-yer?"

"Sommelier."

"Right. Like a wine waitress?"

"I managed a wine program worth 2.8 million dollars."

"Right." He leaned back, chair groaning in protest. "We mostly just do 2-for-1 happy hour wines here. Barefoot, Sutter Home. Nothing fancy."

"I can do that."

"It's just that..." He set down her resume, fingers leaving wrinkles on the paper.

"You're kinda overqualified. These types usually don't last. You know, people who think they're above the work."

"I'm not above anything."

"See, but why? What's someone with all this—" he waved at her resume like it mystified him. "—doing applying at Applebee's?"

She kept her face neutral, the same expression she'd worn while Tony explained that wine was "gay".

"I'm... between positions."

"You get fired?"

"I—"

"'Cause I called your last reference. Tony? Says you walked out mid-shift. That true?"

Her mouth went dry. Of course Tony would say that. Make it sound like she'd just abandoned ship instead of—

"There were circumstances."

"Fuck someone you weren't supposed to?"

She blinked, caught off guard. "What?"

"Screw the boss? Bang a regular's husband?"

The words hit like cold water. Her face must have shown something because he smirked like he caught her red-handed.

"Look," Cole said, tossing her resume in the trash while she watched. "I don't need the drama, and we're not hiring."

————————————————————————

The manager at Olive Garden didn't even look at her resume, just looked her up and down and asked when she could start.

The training was held in a back room that smelled like breadsticks and paper.

The uniform was suffocating polyester that trapped heat against her skin.

Her manager—Pauly—made her spin around in it. Telling her she "filled it out nicely".

"Alright, so we have eight wines," Pauly said, holding the laminated list like it was sacred text. "The Sutter Home White Zin is our biggest seller. Super refreshing."

She'd once spent forty minutes explaining to a table why White Zinfandel wasn't a white wine.

"Got it," she said.

"Now the Riunite Lambrusco - it's Italian, so it's fancier. Pairs with the Tour of Italy."

She'd sourced a '97 Lambrusco di Sorbara for Le Cinq once. The sommelier community had lost their minds when she'd gotten it stateside.

"And our house red is Barefoot Merlot. Good with anything red sauce."

Barefoot. The wine you could get in a box at CVS.

"Any questions?" Pauly asked, leaning in close enough that she could smell his Axe body spray mixing with the heat of his breath.

Yes. How many shifts would it take before she stopped being able to pair properly?

"None."

During the Sunday lunch shift, a woman in a bedazzled "It's Wine O'Clock Somewhere" T-shirt sent back her glass of Barefoot because it was "corked."

She stared at the screw top bottle. At Le Cinq, she'd have educated her. Explained TCA couldn’t come from screw tops. Explained reduction versus bottle shock. She would've made it a teaching moment.

"I'll get you a new glass," she said instead, pouring from the same bottle in the server station. Added an ice cube. The woman loved it.

This was who she was now. Someone who put ice cubes in wine.

The next shift, Pauly waved her over before she could clock in.

Two other servers were nearby, Ashley rolling silverware, the new guy whose name she'd already forgotten folding napkins.

He circled her slowly. She kept her eyes ahead.

"Turn around."

She turned. He tugged at the hem of her shirt, fingers grazing her hip. "Supposed to hit right here." His hand lingered at her waistband. "And it needs to be tucked tighter."

The other servers kept working, but she could feel them watching as she turned.

She stepped back. "Got it."

"Here, let me—" He reached for her shirt.

"I got it." She tucked it herself, heart racing slightly.

He watched her do it. "Better. Can't have you looking sloppy out there."

That night, she googled "wine director positions in Chicago." Then closed the laptop knowing everyone in the industry would know exactly why she wasn't at Le Cinq anymore. Mania wasn't exactly a secret in their incestuous world.

Pauly slid up to her during the next shift. "You never smile."

"I'm smiling right now."

"That's not a smile, that's like… angry."

He wasn't wrong. Everything about her felt angry. Angry at the guests ordering Barefoot. Angry at Pauly who kept calling her "babe." Angry at herself for destroying her dream job.

"Maybe you just need someone to make you smile. How about dinner?"

She immediately thought about that dinner with Jake that had cost her her job.

"I have a boyfriend."

"Yeah? Where's he work?"

"The Bear."

Sure. Why not.

His eyes flicked down to her hand. "No ring though."

————————————————————————

Three weeks in, she was dead on her feet after working a double, her uniform stuck to her skin making her feel suffocated. The piss stairs felt insurmountable right now. She was risking the main entrance.

She turned the corner and—fuck. Of course.

His chef's coat hung open, the undershirt beneath stained and coated in sweat. His knuckle was split. Dried blood between the cracks. He kept rotating his shoulder like something was out of place.

They reached the stairs at the same time. Both standing there, both too exhausted for whatever this was.

"Go ahead," he said, stepping back.

"Long night?" she asked, not even sure why.

"It’s— yeah, fuck— It's fine."

"You're bleeding."

"Yeah— It’s fine Y/n.”

She should just go. Walk past him. But something made her stay.

"Rough service?"

His laugh was all air. "Yeah, no, it was—it was good. Great actually." He ran a hand through his hair. "No one walked out on their check, so."

She felt something hot behind her eyes.

Fuck him.

"Goodnight, Carmen."

She took the stairs, counting each step methodically. The polyester uniform making a swishing sound as she walked.

She got to the fifth step before anger won out.

She came back down, yanking her wallet out of her purse, counted out three twenties. Her grocery money, but whatever. She'd figure it out.

"Here."

Carmen looked at the bills in her extended hand.

"What's this?"

"The money.” She grabs his hand placing the crumpled bills in his fist. "Since it's such a big fucking deal."

He laughed in disbelief, breath fogging the air. "You— you're fucking serious right now?"

"Don't want you to think I'm the kind of person who walks out on a check."

"The check was three hundred."

"You'll get the rest when I have it."

The silence stretches. His eyes roaming over her uniform, trying to place it.

Something shifted in his expression—pity?

Fuck him if that was pity.

"Unless you need it all tonight? Rent due?"

He exhaled sharply through his nose.

"Fuck you." He closes his fist around the bills. "Yeah. Pay off your debt."

"Every fucking penny, chef," she said, jabbing his fist for emphasis. “Count on it."

————————————————————————

First paycheck barely covered groceries and her lamictal. Second paycheck, rent was still short. She was calculating if she could make it to the third when she heard it.

That laugh she instantly recognized. Bright and cutting, like nails on chalkboard. Her stomach dropped.

Margot from Le Cinq.

They'd once torn apart a new somm at Spiaggia together. What was his name? Daniel? David? He quit the next day.

She was sitting with two others. She recognizes Elias, tje somm from Alinea. The other she doesn't know but he has that look—industry, successful, probably runs somewhere important.

They're clearly here ironically, probably already tipsy from somewhere better.

She considers running. But Pauly's watching her, and she needs this job.

She approaches the table. Margot's eyes go wide, then narrow with obvious delight.

"Oh my god. Y/N?" That voice, pitched to carry.

"Hi Margot."

"What are you... you work here?" Like it's impossible. Her eyes tracking over her like it's a costume party.

"Can I start you with something to drink?"

"Holy shit, you do." Margot covers her mouth with her hand.

Elias is looking at her uniform like it physically hurts him.

He's cataloging every thread of the polyester. He was going to tell someone about it later. ‘Did you hear about Y/N? Yeah, she's at Olive Garden now. No, I'm serious.’

"We have eight wines—"

"Eight!" Margot laughs, turning to the third guy. "She's a real somm— no joke! She used to manage Le Cinq's wine."

The third guy pipes up. "Wait, the one who destroyed the walk-in?"

"During service," Margot confirms, "Cost them forty grand in product. Plus the repair."

Her shoulders curled inward.

"Would you like to hear about our wine selection?"

"Please!” Margot leans forward eagerly, chin on her fist, eyebrows raised. "Tell us about your wine.”

She recites the eight wines like she's reading a hostage statement. Barefoot. Sutter Home. Riunite Lambrusco—"it's Italian, so it's fancier."

Elias winces, Margot's recording on her phone.

Don't look at it. If you look at it, you'll grab it. If you grab it, you'll lose this job too.

"A Spumante?" The third guy asks.

"It's actually a frizzante, not a spumante, so the bubbles are—" She catches herself. Margot's smirk is spreading, phone still recording. "It pairs well with the Tour of Italy."

"I'm sure it does."

Margot orders the Barefoot Moscato, tells the others they have to get it for the full experience.

"This is incredible. You're so humble now."

She takes their order. Brings the wines. Margot makes her explain the difference between the marinara and the meat sauce.

When she brings the check, Margot leaves a $100 dollar tip on their eighty dollar bill. Writes "for the excellent service!!” with a smiley face on the receipt.

————————————————————————-

Her legs shake as she pushes through the back door. There's a milk crate from someone's smoke breaks.

Thank god, because her knees buckle as soon as she reaches it.

Her pants immediately absorb the water from the earlier rain still on it, cold spreading across her thighs.

She tries to breathe but it comes out wrong—a sob, ugly and raw. Her hands shake as she presses them hard against her eyes, the sobs keep coming, each one ripping through her chest.

The door bangs open behind her.

"Jesus, you okay?"

Pauly. Of course. She can hear his Nikes splashing through the puddles on the pavement, getting closer.

"Hey, it's okay."

He crouches in front of her.

"Those customers give you shit?"

She tries to answer but another sob breaks through.

"Aw, come on. Some customers are dicks. You can't take it personal." His hand lands on her shoulder, squeezing.

"You're too pretty to cry over this shit."

"I'm fine," she manages, trying to shift away, but she's trapped between him and the brick wall, the dumpster boxing her in on the left.

He shifts closer, she can smell the garlic on his uniform, the Axe underneath.

"You're not fine." His voice drops, growing softer. His other hand lands on her knee. "You're shaking."

She didn't even realize it.

"Got some weed in my car. We could go somewhere, blow some steam off." His thumb strokes her leg. “Come on, I'll take care of you tonight."

The kitchen door is closed. She can't hear the dishwashers anymore— they must've finished. The parking lot is empty except for his Honda with mismatched doors and a van that's been there since yesterday.

His hand slides up from her knee, fingers spreading wide across her thigh. The polyester is so thin she can feel his pinky nail scratching against the fabric. His thumb finds the inner seam of her pants, pressing along it, forcing her legs to open slightly. She can see the kitchen timer tattoo on his wrist, the one he got in culinary school—he'd told that story during training, proud of lasting two semesters before dropping out.

His thumb is still moving. The same thumb that counts the bills from the register, that high-fives the bussers, that jabs at her when he wants her to take tables. She feels sick.

"I've seen girls like you before. Comin’ in all motivated, thinking they're gonna move up fast."

He shifts his weight, blocking more of her view of the door.

"But you know who actually moves up fast? Gets the good sections?" His hand squeezes her thigh.

“Team players.”

Her mouth runs dry.

"My boyfriend's coming to pick me up," she blurts out.

His hand stills. "From The Bear?"

She nods. "He's already close."

She crosses her legs, forcing his hand to move.

He nods. "Lucky guy."

He stands, knees cracking. "Must really love you to be picking you up at midnight."

"Yeah."

He stands there for a moment, staring down at her. She holds her breath.

"See you tomorrow, Y/N."

She watches him disappear inside, counting to 10 before her body unlocks. Her hands shake as she fumbles for her phone then drops it, the screen cracking against the pavement.

She calls an Uber, not caring how she couldn't afford it, she just wanted to get home.

She keeps glancing over her shoulder while waiting for it.

Three minutes away.

How the fuck long was three minutes?

[tzifron]

The AI jobs crisis is here, now

It's not coming, it has already arrived.

I’ve been rereading the late anthropologist David Graeber’s Bullshit Jobs, which persuasively makes the case that the corporate world is happy to nurture inefficient or wasteful jobs if they somehow serve the managerial class or flatter elites—while encouraging the public to harbor animosity at those who do rewarding work or work that clearly benefits society. I think we can expect AI to accelerate this phenomenon, and to help generate echelons of new dubious jobs—prompt engineers, product marketers, etc—as it erodes conditions for artists and public servants.

A common refrain about modern AI is that it was supposed to automate the dull jobs so we could all be more creative, but instead, it’s being used to automate the creative jobs. That’s a pretty good articulation of what lies at the heart of the AI jobs crisis. Take the former Duolingo worker who was laid off as part of the company’s pivot to AI.

“So much will be lost,” the writer told me. “I was a content writer, I wrote the questions that learners see in the lessons. I enjoyed being able be creative. We were encouraged to make the exercises fun.” Now, consider what it’s being replace with, per the worker:

“First, the AI output is very boring. And Duolingo was always known for being fun and quirky. Second, it absolutely makes mistakes. Even on things that you would think it could get right. The AI tools that are available for people who pay for Duolingo Max often get things wrong—they have an ‘explain my mistake’ tool that often will suggest something that’s incorrect, sometimes the robot voices are programmed to speak the wrong language.”

This is just a snapshot, too. This is happening, to varying degrees, to artists, journalists, writers, designers, coders—and soon, perhaps already, as Thompson’s story points out, it could be happening to even more jobs and lines of work.

Now, it needs to be underlined once again that generative AI is not yet the one-size-fits-all agent of job replacement its salesmen would like it to be—far from it. A recent SalesForce survey reported on by the Information show that only one-fifth of enterprise AI buyers are seeing good results, and that 61% of respondents report a disappointing return on investment for AI or even none at all.

Generative AI is still best at select tasks that do not require consistent reliability—hence its purveyors taking aim at art and creative industries. But all that’s secondary. The rise of generative AI, linked as it is with the ascent to power of the American tech oligarchy, has given rise to a jobs crisis nonetheless.

We’re left at a crossroads where we must consider nothing less than what kind of jobs we want people to be able to do, what kind of work and which institutions we think are important as a society, and what we’re willing to do to protect them—before the logic of generative AI and the jobs crisis it has begotten guts them to the bone, or devours them altogether.

[shituationist]

Finally got the cursor demo at work. Assuaged some of my fears about Total Automation of Computer Touching, and honestly, it's funny how much of it would amount to copying and pasting from documentation. It's absolutely an exercise in deskilling. You're basically trying to get it to retrieve code from the model weights based on the probabilistic massaging of the context window. You don't have to write terse code, instead you can write way longer natural language prompts. Some of that is a one time cost (basically centralizing the knowledge of your code base into "skills" markdown) but in the time it took us to write a pretty simple python report we could have written two without the aid of the program. And god help us if we ever make any changes that render the skill markdown out of date. The entire exercise becomes one of maintaining documentation to make the probabilistic sequence extruder more probable.

I think the fretting about this putting programmers out of work is only warranted in the sense that demand for programmers is falling, and not because these tools can do the jobs of programmers without programmers, but because so much software is just superfluous. Already before LLMs were commercialized, we were running up against the industry running out of shit to sell. The main benefit of using software versus, say, circuits or pen and paper, is that it lets you handle shit like data processing at scale better. If you have a spreadsheet with 10,000 rows and you need to plot those data points along with a linear regression of those points, you'd better use a computer program, because doing it by hand would be tedious, time consuming, and error prone. But if you just need to send a dozen invoices a day? Track the work in progress in your autoshop? Pen and paper is fine, and it never goes offline when a data center explodes. There's a reason that "laggard industries" like automotive repair shops are so hard to sell software too. Not only do they not need data processing at scale (the main benefit of using software), most of what they actually need for this comes with a basic enterprise Windows license, or packaged for free with any Linux distro (srsly, LibreOffice Calc is just fine for almost all business applications).

It's gonna suck if our jobs basically become documentation curation plus prompt fondling (and at least for now this doesn't apply to large and unwieldy code bases, and for tasks not well represented in the training corpora), but maybe an unintended benefit of that is that software loses much of its social prestige, and with that, the dumb opinions of software engineers and software capitalists. I've always thought it was obscene that software engineers make as much as we do, with our minimal professional standards and ethics. Much of that is the result of our industry's vague futurist promises, but we are at the end of the hype cycle for software in general, and with greater knowledge of the social costs of our whole industry. The world needs less software, if anything.

[quartz-components]

Why Raspberry Pi Continues to Power Innovation? 🚀

Innovation doesn't always begin in a billion-dollar laboratory. Sometimes, it starts on a desk, in a classroom, or at a workbench—with a credit-card-sized computer called the Raspberry Pi.

Since its introduction, Raspberry Pi has transformed the world of embedded computing by making technology more accessible to students, hobbyists, educators, startups, and professional engineers alike. What started as an educational platform has evolved into one of the most trusted computing solutions for prototyping, automation, AI applications, industrial control systems, and edge computing.

One of the biggest reasons behind Raspberry Pi's popularity is its incredible flexibility. Whether you're just learning Python or developing a commercial product, the same platform can adapt to your needs. Its extensive GPIO interface allows seamless integration with sensors, motors, displays, cameras, communication modules, and thousands of electronic components, making it suitable for projects of every scale.

What Can You Build?

The possibilities are nearly endless. Raspberry Pi is widely used for:

🤖 Autonomous robots and robotic arms

🏠 Smart home automation systems

🌱 Agricultural monitoring and smart farming

📷 AI-powered computer vision applications

🌐 IoT gateways and remote monitoring solutions

🎮 Retro gaming consoles and media centers

📊 Industrial data acquisition systems

🛰️ Edge computing and machine learning deployment

🎓 STEM education and electronics training

💻 Personal servers, NAS, and cloud storage projects

Why Developers Love Raspberry Pi

Unlike many development platforms, Raspberry Pi combines powerful hardware with an enormous global community. Millions of developers continuously contribute tutorials, open-source software, libraries, and project ideas, making it easier than ever to solve problems and accelerate development.

Some of its key advantages include:

✔ Compact yet powerful computing platform

✔ Excellent Linux operating system support

✔ Python, C/C++, Java, Node.js, and many other programming languages

✔ Extensive GPIO for hardware interfacing

✔ Camera and display compatibility

✔ Wi-Fi, Bluetooth, Ethernet, and USB connectivity

✔ Large ecosystem of HATs, accessories, and expansion boards

✔ Reliable platform for education, research, and industrial applications

Beyond Learning

Today, Raspberry Pi is no longer just a board for beginners. It powers digital signage, factory automation, AI inference devices, kiosks, surveillance systems, laboratory equipment, environmental monitoring stations, and countless commercial products around the world. Businesses rely on it for rapid prototyping, while educators use it to inspire the next generation of innovators.

Its affordability, reliability, and versatility have made it one of the most influential pieces of hardware in modern electronics.

Whether you're designing your first LED blinking project, building an intelligent robot, deploying an IoT network, or creating an AI-powered edge device, Raspberry Pi provides the perfect foundation to turn ideas into reality.

Every breakthrough begins with curiosity—and sometimes, all it takes is a tiny computer to build something extraordinary.

✨ What are you creating with Raspberry Pi? We'd love to hear about your latest project, your biggest challenge, or the idea you're bringing to life.

[mariacallous]

The Math on AI Agents Doesn’t Add Up

A research paper suggests AI agents are mathematically doomed to fail. The industry doesn’t agree.

The big AI companies promised us that 2025 would be “the year of the AI agents.” It turned out to be the year of talking about AI agents, and kicking the can for that transformational moment to 2026 or maybe later. But what if the answer to the question “When will our lives be fully automated by generative AI robots that perform our tasks for us and basically run the world?” is, like that New Yorker cartoon, “How about never?”

That was basically the message of a paper published without much fanfare some months ago, smack in the middle of the overhyped year of “agentic AI.” Entitled “Hallucination Stations: On Some Basic Limitations of Transformer-Based Language Models,” it purports to mathematically show that “LLMs are incapable of carrying out computational and agentic tasks beyond a certain complexity.” Though the science is beyond me, the authors—a former SAP CTO who studied AI under one of the field’s founding intellects, John McCarthy, and his teenage prodigy son—punctured the vision of agentic paradise with the certainty of mathematics. Even reasoning models that go beyond the pure word-prediction process of LLMs, they say, won’t fix the problem.

“There is no way they can be reliable,” Vishal Sikka, the dad, tells me. After a career that, in addition to SAP, included a stint as Infosys CEO and an Oracle board member, he currently heads an AI services startup called Vianai. “So we should forget about AI agents running nuclear power plants?” I ask. “Exactly,” he says. Maybe you can get it to file some papers or something to save time, but you might have to resign yourself to some mistakes.

The AI industry begs to differ. For one thing, a big success in agent AI has been coding, which took off last year. Just this week at Davos, Google’s Nobel-winning head of AI, Demis Hassabis, reported breakthroughs in minimizing hallucinations, and hyperscalers and startups alike are pushing the agent narrative. Now they have some backup. A startup called Harmonic is reporting a breakthrough in AI coding that also hinges on mathematics—and tops benchmarks on reliability.

Harmonic, which was cofounded by Robinhood CEO Vlad Tenev and Tudor Achim, a Stanford-trained mathematician, claims this recent improvement to its product called Aristotle (no hubris there!) is an indication that there are ways to guarantee the trustworthiness of AI systems. “Are we doomed to be in a world where AI just generates slop and humans can't really check it? That would be a crazy world,” says Achim. Harmonic’s solution is to use formal methods of mathematical reasoning to verify an LLM’s output. Specifically, it encodes outputs in the Lean programming language, which is known for its ability to verify the coding. To be sure, Harmonic’s focus to date has been narrow—its key mission is the pursuit of “mathematical superintelligence,” and coding is a somewhat organic extension. Things like history essays—which can’t be mathematically verified—are beyond its boundaries. For now.

Nonetheless, Achim doesn’t seem to think that reliable agentic behavior is as much an issue as some critics believe. “I would say that most models at this point have the level of pure intelligence required to reason through booking a travel itinerary,” he says.

Both sides are right—or maybe even on the same side. On one hand, everyone agrees that hallucinations will continue to be a vexing reality. In a paper published last September, OpenAI scientists wrote, “Despite significant progress, hallucinations continue to plague the field, and are still present in the latest models.” They proved that unhappy claim by asking three models, including ChatGPT, to provide the title of the lead author’s dissertation. All three made up fake titles and all misreported the year of publication. In a blog about the paper, OpenAI glumly stated that in AI models, “accuracy will never reach 100 percent.”

Right now, those inaccuracies are serious enough to discourage the wide adoption of agents in the corporate world. “The value has not been delivered,” says Himanshu Tyagi, cofounder of an open source AI company called Sentient. He points out that dealing with hallucinations can disrupt an entire work flow, negating much of the value of an agent.

Yet the big AI powers and many startups believe these inaccuracies can be dealt with. The key to coexisting with hallucinations, they say, is creating guardrails that filter out the imaginative bullshit that LLMs love to produce. Even Sikka thinks that this is a probable outcome. “Our paper is saying that a pure LLM has this inherent limitation—but at the same time it is true that you can build components around LLMs that overcome those limitations,” he says.

Achim, the mathematical verification guy, agrees that hallucinations will always be around—but considers this a feature, not a bug. “I think hallucinations are intrinsic to LLMs and also necessary for going beyond human intelligence,” he says. “The way that systems learn is by hallucinating something. It’s often wrong, but sometimes it's something that no human has ever thought before.”

The bottom line is that like generative AI itself, agentic AI is both impossible and inevitable at the same time. There may not be a specific annum that will be looked back upon as “the year of the agent.” But hallucinations or not, every year from now on is going to be “the year of more agents,” as the delta between guardrails and hallucinations narrows. The industry has too much at stake not to make this happen. The tasks that agents perform will always require some degree of verification—and of course people will get sloppy and we’ll suffer small and large disasters—but eventually agents will match or surpass the reliability of human beings, while being faster and cheaper.

At that point, some bigger questions arise. One person I contacted to discuss the hallucination paper was computer pioneer Alan Kay, who is friendly with Sikka. His view is that “their argument was posed well enough to get comments from real computational theorists.” (A statement reminiscent of his 1984 take on the Macintosh as “the first personal computer good enough to be criticized.”) But ultimately, he says, the mathematical question is beside the point. Instead, he suggests people consider the issue in light of Marshall McLuhan’s famous “Medium is the message” dictum. “Don’t ask whether something is good or bad, right or wrong,” he paraphrases. “Find out what is going on.”

Here’s what’s going on: We may well be on the cusp of a massive automation of human cognitive activity. It’s an open question whether this will improve the quality of our work and our lives. I suspect that the ultimate assessment of that will not be mathematically verifiable.

[quashtech]

Data Science Course in Thane: Fees, Duration, Syllabus, Placement & Career Guide (2026)

Introduction

Data science is one of the fastest-growing fields in the technology sector. Industries such as finance, healthcare, e-commerce, logistics, and IT rely heavily on data to make strategic business decisions. Because of this demand, many students and working professionals are exploring data science courses to build analytical and technical skills.

For learners in Thane, choosing the right program involves understanding the course syllabus, learning format, career opportunities, and practical exposure. This guide explains what a modern data science course includes and how to select the right training program.

What Is a Data Science Course?

A data science course in Thane teaches students how to process large volumes of data and extract meaningful insights. The training usually combines programming, statistics, and data visualization.

Students typically learn how to:

Collect and clean raw data

Analyze patterns and trends

Build predictive models

Present insights using dashboards and reports

These skills help organizations make better decisions and improve operational efficiency.

Syllabus and Tools Learned in a Data Science Course

A strong syllabus is one of the most important factors when selecting a data science program. Industry-oriented courses generally cover the following areas.

Python Programming

Python is the most widely used programming language in data science. Students usually work with popular libraries such as:

NumPy

Pandas

Matplotlib

Scikit-learn

These tools are used for processing large datasets and performing statistical analysis.

Statistical and Data Analysis

A solid understanding of statistics is essential for interpreting data. Common topics include:

Probability

Regression analysis

Hypothesis testing

These concepts help analysts identify patterns and make predictions from data.

Data Visualization

Visualization tools convert complex datasets into clear and understandable insights. Students often learn tools such as:

Power BI

Tableau

These tools help create dashboards and visual reports for business decision-making.

Machine Learning Basics

Many data science courses introduce machine learning algorithms that help build predictive models for tasks such as forecasting, automation, and pattern detection.

Data Science Course Fees and Duration in India

In India, the cost of a data science course varies depending on the learning mode and training level.

Typical course durations include:

3–6 months for beginner or intermediate programs

6–12 months for advanced or comprehensive courses

Course fees depend on several factors:

Curriculum coverage

Trainer experience

Project exposure

Career support

Instead of considering cost alone, learners should also evaluate the quality of training and practical learning opportunities.

Online vs Offline Data Science Course: Which Is Better?

Both online and classroom learning offer unique advantages.

Online Data Science Courses

Online courses are flexible and convenient for students who cannot attend physical classes.

Benefits include:

Flexible learning schedule

Access to recorded sessions

Self-paced study options

Offline Classroom Courses

Classroom training provides direct interaction with instructors and classmates.

Advantages include:

Live problem-solving sessions

Instant doubt clarification

Hands-on lab practice

Many institutes now offer hybrid learning models that combine online flexibility with classroom interaction.

Placement-Focused Data Science Courses

Placement-oriented programs focus on job readiness and practical learning. Students build industry-relevant skills such as:

Data analysis using Python and SQL

Data visualization and reporting

Applying machine learning models to real problems

Presenting insights through dashboards and reports

Practical projects and portfolio development are important for preparing students for interviews.

Best Data Science Courses for Beginners

Beginners should choose courses that start with basic concepts and gradually move toward advanced topics.

A beginner-friendly course usually includes:

Python fundamentals

Data cleaning and preprocessing

Data visualization tools

Introduction to machine learning

Hands-on projects are essential because they help learners apply theoretical knowledge to real-world scenarios.

Data Science Courses for Working Professionals

Working professionals often prefer online or weekend programs that allow them to balance learning with their jobs.

Professional certification programs typically focus on:

Real-world business case studies

Advanced data analysis tools

Portfolio development

Career transition guidance

These programs help professionals upgrade their skills and move into analytics or data science roles.

Choosing the Right Data Science Training Institute

Selecting the right training institute requires evaluating several important factors:

Industry-relevant curriculum

Practical project experience

Experienced trainers

Transparent course structure

Career guidance and counseling

For learners considering data science training in Thane, institutes such as Quastech IT Training and Placement Institute are often mentioned for their focus on practical IT training and structured learning programs.

Career Guide After Completing a Data Science Course

After completing a professional data science program, learners can explore roles such as:

Data Analyst

Junior Data Scientist

Reporting Analyst

Business Analyst

With experience, professionals can advance to roles such as:

Machine Learning Engineer

Analytics Consultant

Data Science Lead

Final Thoughts

A best data science course in Thane can open the door to many career opportunities in today’s digital economy. However, selecting the right training program is essential.

Learners should focus on curriculum quality, practical exposure, and industry relevance, rather than only the popularity of the course.

For students and professionals in Thane, structured training combined with consistent practice can help build a strong foundation in data science and support long-term career growth in the technology industry.