The anti-AI thread

Page 11 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Kaido

Elite Member & Kitchen Overlord
Feb 14, 2004
51,576
7,244
136
Nestlé to cut 16,000 jobs globally amid cost reductions and AI concerns:


Most of the layoffs – about 12,000 – will affect white-collar professionals as Nestlé targets “operational efficiency,” including by automating processes and using shared services, the company behind such brands as KitKat and Nesquik said in a statement.

Nestlé says on its website that it uses AI in a number of functions, including research and development. In its last annual report, it also said it employs automation and advanced analytics in promotional activities, such as work on discounts and in-store displays.

 

Kaido

Elite Member & Kitchen Overlord
Feb 14, 2004
51,576
7,244
136
  • Like
Reactions: marees and Racan

Kaido

Elite Member & Kitchen Overlord
Feb 14, 2004
51,576
7,244
136
AI Is Killing Wikipedia’s Human Traffic:


In a blog post published today, Marshall Miller, the foundation’s senior director of product, said Wikipedia’s human visits are down about 8% over the past few months compared to the same period in 2024.

 
  • Like
Reactions: marees

mikeymikec

Lifer
May 19, 2011
21,007
16,259
136
AI Is Killing Wikipedia’s Human Traffic:





All part of the plan. In future, please insert money into AI for answers.
 
  • Like
Reactions: KMFJD

mikeymikec

Lifer
May 19, 2011
21,007
16,259
136
I'm generally of the opinion that AI in its current form (aside from some very niche applications) is going to fail hard, I'm in agreement with the market suspicion that it's a bubble that's going to burst (and IMO that burst is a good thing because the entire premise was flawed), however I'm increasingly of the opinion that those in the market driving for this sort of AI are doing so on a premise that makes sense: People want easy* answers, and AI can deliver on that.

* - please note I just used the word "easy", this does not imply a trifecta of say: easy, quick and accurate.

There's a quote from Incredibles 2 that I think is very appropriate:
"People want ease. People will trade quality for ease every time. It may be crap, but it's convenient!"
 
Last edited:
  • Like
Reactions: Racan and KMFJD

[DHT]Osiris

Lifer
Dec 15, 2015
17,376
16,655
146
I'm generally of the opinion that AI in its current form (aside from some very niche applications) is going to fail hard, I'm in agreement with the market suspicion that it's a bubble that's going to burst (and IMO that burst is a good thing because the entire premise was flawed), however I'm increasingly of the opinion that those in the market driving for this sort of AI are doing so on a premise that makes sense: People want easy* answers, and AI can deliver on that.

* - please note I just used the word "easy", this does not imply a trifecta of say: easy, quick and accurate.

There's a quote from Incredibles 2 that I think is very appropriate:
"People want ease. People will trade quality for ease every time. It may be crap, but it's convenient!"
I agree with you that there is a hype bubble surrounding AI. It's also true though that it's not going away. Inside that bubble is a core of molten iron, I don't know how big it is compared to the entire bubble but you aren't dislodging it at this point.

Large swathes of human productivity will be run by AI very soon (depending on your version of very).
 
  • Like
Reactions: KMFJD

marees

Golden Member
Apr 28, 2024
1,807
2,426
96
The below is my opinion too

Nightcap: What about the argument that ChatGPT, while not perfect, is capable of doing some low-level grunt work could increase productivity?

Garran:
There are certain bullsh*t jobs out there — some parts of management, consultancy, jobs where people don’t check if you’re getting it right or don’t know if you’ve got it right. So you can argue that you can replace bullsh*t with bullsh*t, and, yes, OK, I’m prepared to accept that you probably can, but that doesn’t really make it more broadly useful.


However I think in order to recover the sunk costs, everything in this world will be enshittified so that AI can replace them then
 

Kaido

Elite Member & Kitchen Overlord
Feb 14, 2004
51,576
7,244
136
"People want ease. People will trade quality for ease every time. It may be crap, but it's convenient!"

The reason for this is simple:

* The brain is an energy gatekeeper

We all live with an internal 2-party system:

1. Our mind (conscious choices)
2. Our brain (energy manager)

Our mind works off iteration (step execution, ie. actually doing things), whereas our brain works off ideas. Those ideas are subject to being controlled by our "energy perspective" (i.e. how much energy we have available at the time to cope with imagining doing them). That's why we'll put off doing the laundry all week (idea), when in reality it's a simple 10-minute job: (steps)

1. Dump dirty clothes in washer & run (2 minutes)
2. Swap washer to dryer an hour later (1 minute)
3. Fold one batch of laundry (~7 minutes max for a large load)

This is called the HCEG (Hot-Cold Empathy Gap). Basically:

1. In our cold logical state, we make confident plans
2. Later, in our hot emotional state, those plans fall apart, because we failed to take into account the variable energy states that drive mood

This applies to literally everything:

* Signing up for a gym membership & then not wanting to go
* Trying to quit addictions like smoking, but then caving when the urge hits
* Buying salads at the grocery store, but then getting drive-thru fast food on the way home

As an energy filter, the brain does the following behavior by default:

* Takes the path of least resistance

This, however, is misunderstood! There are 2 key factors in understanding that reality:

1. The brain is an energy scrooge...but not across the board! It will allocate unlimited energy for things you WANT to do (such as play video game for 6 hours straight until midnight), but then zap your energy to do the dishes lol.

2. The path of least resistance is NOT the most simple path, unlike Occam's Razor! It is the path that requires the LEAST amount of energy expenditure, which means taking the path we already know because it requires ZERO energy to figure out! This is why habits & addictions rule our lives & why it's so hard to actually change! It's not about desire, but rather, ENERGY!

This is what makes what I call the "lull of complacency" so dangerous: when things are "good enough", we tend to buy in & quit! Sample situations:

1. Getting to a point where we are comfortable in life. 85% of people hate their job

2. Tesla Full-Self-Driving has been responsible for hundreds of crashes. The system is VERY good for what it is , but requires constant driver vigilance because it's just good enough to lull people into that complacency trap. Without exception, every single Tesla driver I know plays on their phone in FSD mode.

3. I personally consider food to be the #1 preventatable killer in America. We are 40%obese, 50% diabetic, and 73% overweight. 73% of the American supply chain is ultra-processed food!

4. ChatGPT hallucinates...but, like fast food, it is "good enough" for most people!

There will ALWAYS be a HUGE market for convenience because human beings operate off emotion-based motivation (tactical living, based on mood) instead of commitment-based motivation (strategic living, based on systems). This means that there are ENORMOUS opportunities for those who are willing to design & sell convenience! For example, supermarket checkout aisle impulse purchases like gum, mints, and candy bars is a $6 BILLION dollar market:


Nearly nine out of ten American households visited a McDonald's in the past year. They do $26 BILLION in sales worldwide annually:


Just look at their effort-reduction tactics:

1. Can get food from walking in, the rive-thru, parking lot delivery via the app, and courier delivery
2. Same reliable flavor every time, which appeals to our automaticity (you never know how homemade dinner will turn out!)
3. Affordable to the majority of Americans, despite recent inflation
3. No effort to cook required
4. No effort to clean required (no dishes!)
5. Decision fatigue reduction (3 entree options: beef, chicken, fish. 3 dessert options: ice cream, pies, cookies).

ChatGPT is used by more than 750 million people, which is nearly 10% of the world’s population! At the current growth rate, it will surpass Google for the number of daily searches by next year:


I have NEVER seen a technology adopted so fast by the average non-technical consumer, not even smartphones! Here's some Internet nostaglia for you:


And today, 17 years later:

 

nakedfrog

No Lifer
Apr 3, 2001
62,800
18,995
136
Why the fuck do we need AI to verify a picture a human is taking of another human at the DMV? What's the point?


(That article is mostly just repackaging part of a Wired article, which takes a subscription, it describes the difficulties some other folks have with facial recognition as well)

 
  • Like
Reactions: marees

Kaido

Elite Member & Kitchen Overlord
Feb 14, 2004
51,576
7,244
136
AI to verify a picture

Watch out for the Meta Ban Wave:





Facebook started a ban for various good reasons over summer (harmful content etc.), but then businesses & non-offenders also got perma-banned...people with 15+ year old accounts, Instagram accounts, Quest headset accounts & all associated purposes, small biz pages, etc. For the past few weeks, there's been an appeal loop where people can NEVER get back in. It seems to be a mix of an overzealous AI auto-moderator, a faulty Meta update, and the AWS downtime that has gotten worse. There are rumors that millions of accounts are just completely gone:





It's easy to brush it of as "good, leave Facebook!"), but deplatforming can be HUGELY isolating for people, especially disabled people, as well as business-reducing for so many small businesses. I have friends who rely on FB & IG for selling their services like baking, nails, car detailing, etc. Cambodia in particular has so many bans that the government had to get involved due to the account-restore scams going on:


This is the future, right now, today:


1. Your whole life & business engrained on social media
2. AI ban with no human intervention
3. ¯\_(ツ)_/¯

1761015500553.png
 

Kaido

Elite Member & Kitchen Overlord
Feb 14, 2004
51,576
7,244
136
Of ChatGPT's 800 million users, just 5% pay:


OpenAI has other theoretical revenue streams, like charging a commission for items purchased through ChatGPT e-commerce integrations and ads – something CEO Sam Altman initially dismissed but now is considering.

Automated e-commerce could be a thing, someday. Stranger things have happened. Ad support, however, has proven difficult for rival Perplexity, which recently paused accepting new advertisers to rethink its revenue plan.

OpenAI's platforms account for about 80 percent of all web traffic for generative AI tools, representing 190 million of the 240 million average daily visits, according to SimilarWeb data [PDF] published in May.

OpenAI's path to profitability is easier said than done – make a product so compelling that people will pay for it. We're not there yet. ®
 

Kaido

Elite Member & Kitchen Overlord
Feb 14, 2004
51,576
7,244
136
1984 stuff:

1. Most AI is "Software as a Service"

2. Unwanted changes can happen (re: ChatGPT v5)

3. Unannounced restrictions & bans can happen with vague limits

 

Kaido

Elite Member & Kitchen Overlord
Feb 14, 2004
51,576
7,244
136
Watch out for the Meta Ban Wave:





Facebook started a ban for various good reasons over summer (harmful content etc.), but then businesses & non-offenders also got perma-banned...people with 15+ year old accounts, Instagram accounts, Quest headset accounts & all associated purposes, small biz pages, etc. For the past few weeks, there's been an appeal loop where people can NEVER get back in. It seems to be a mix of an overzealous AI auto-moderator, a faulty Meta update, and the AWS downtime that has gotten worse. There are rumors that millions of accounts are just completely gone:





It's easy to brush it of as "good, leave Facebook!"), but deplatforming can be HUGELY isolating for people, especially disabled people, as well as business-reducing for so many small businesses. I have friends who rely on FB & IG for selling their services like baking, nails, car detailing, etc. Cambodia in particular has so many bans that the government had to get involved due to the account-restore scams going on:


This is the future, right now, today:

1. Your whole life & business engrained on social media
2. AI ban with no human intervention
3. ¯\_(ツ)_/¯

View attachment 132326

Medical support & automatic AI moderation:


This summer, my Facebook account was permanently “disabled.”

I had been helping another mother navigate a medical decision for her child, a young adult who is living with a degenerative condition. It wasn’t medical advice. It was empathy, drawn from my own lived experience as a medical mother, a certified coach, and years of teaching courageous communication at a medical school.

Meta AI had flagged the conversation as a violation of Community Standards on child abuse. My words, marked by an algorithm that couldn’t distinguish exploitation from support. My account, and years of advocacy, caregiving, and connection, disappeared overnight.

It took six weeks, multiple appeals, and a friend-of-a-friend inside the company to reach a human being who confirmed it had been a mistake.

By the time my account had been restored, something in me had shifted.

Algorithms don’t understand context or care.
For families like mine (parents navigating rare diseases, disability, or chronic illness) online communities have become lifelines.
This is where we go when the rest of the world sleeps. I can post in a support group in the middle of the night, and find another parent across the country, or across the world, who understands. Someone who doesn’t need the backstory to respond with compassion. Peer-to-peer communities, often hosted on platforms like Facebook and Instagram, have quietly become part of our public health infrastructure. They lower isolation, reduce caregiver stress, and increase engagement with care plans. These same spaces are now being monitored by algorithms that flag “dangerous content.” When an AI system can’t tell the difference between misinformation and a parent sharing fear or uncertainty, it can silence the very support families depend on. When language about fear, prognosis, or end-of-life care is automatically deemed suspect, taken out of context and out of community, we risk losing the capacity to talk about the hardest parts of medicine at all.

When we ban words, we lose people.

There’s a quiet irony here. Medicine already struggles with language: the words we avoid, the silences that form around suffering, disability, and uncertainty.

Now, those silences are being automated.

If algorithms start deciding which stories are safe to tell, we risk losing the spaces where caregivers and families process what cannot be fixed.

These are not peripheral conversations. They are central to healing.

Clinicians need to care about where families are finding support and how those spaces are being shaped. Because if families can’t talk about what scares them online, in spaces built for comfort and connection, they may stop talking about it altogether. Especially in the clinic.

Engagement is everything

Doctors worry about misinformation online. Rightly so. Social media is rife with false expertise and outright fabrication. But the solution isn’t censorship. It’s engagement. Families rarely turn to Facebook because they distrust their doctors. They join because they need to be heard. They want someone to stay with them in the unknowing. When medicine steps out of the conversation, we leave room for fear to grow in silence.

Health care professionals need to be aware of and engaged in these digital spaces. To model, not to monitor. To partner in what respectful, evidence-informed, compassionate dialogue can look like. Doctors, nurses, allied health professionals, and educators can play a vital role in fostering healthy peer-to-peer support networks. The same empathy brought to the bedside can be extended to the comment thread.

Connection cannot be automated. Listening cannot be outsourced.

We need to build softer communities.

As a mother-scholar, I live in the dual worlds of clinical education and clinical navigation. Through my courses and workshops, I remind health care professionals that engagement is not a distraction from professionalism. It’s part of it. I’m rebuilding those softer spaces through my Substack, The Soft Bulletin, and through my coaching work with clinicians and caregivers. I’m not leaving connection behind. I’m rebuilding it. What I want, and what I believe many clinicians and caregivers want, is a softer kind of community. One that values curiosity over compliance, listening over labeling, and conversation over control.

As health care professionals, we must ask: Where are our patients finding connection? What happens when the algorithms that shape those spaces decide their words are dangerous?

If we want to protect mental health, trust, and humanity in medicine, we have to keep talking.

Because engagement is everything. Healing doesn’t happen in isolation. It happens in dialogue. Even, and especially, the hard conversations.
 
  • Like
Reactions: marees