Every "AI" related business idea I've seen prop up recently is people just hooking up a textbox to ChatGPT's API and pretending they're doing something novel or impressive, presumably to cash in on VC money ASAP. The Notion AI is an absolute fucking joke of epic proportions in its uselessness yet they keep pushing it in every newsletter

And a funny personal anecdote, a colleague of mine tried to use ChatGPT4 when answering a customer question (they work support). The customer instantly knew it was AI-generated and was quite pissed about it, so the support team has an unofficial rule to not do that any more.

Comparisons between AI and crypto are horribly misguided IMO.

Is AI overhyped? Sure. However -

AI/ML is creating utility everywhere in our lives - speech to text, language translation, recommendation engines, relevancy ranking in search, computer vision, etc. and seems to be getting embedded in more and more processes by the day.

Crypto never amounted to anything beyond a currency for black market transaction, a vehicle for speculation, and a platform for creating financial scams.

That’s exactly the hype talk that’s going to burst this bubble.

Here’s tech’s dirty little secret. Despite all the screams about automation and universal basic income… the exact numbers where job replacement would show are in the labor productivity numbers. If GDP stays flat or grows while the number of jobs is reduced… bingo… you’d see that number climb.

Productivity has actually stayed flat or gone down over the last 15 years. Despite the fact that we’ve had trillion dollar corporate behemoths now. Despite that fact that we’re enabling a surveillance state Orwell couldn’t imagines. Despite the polarization we see. And teen anxiety going through the roof along teen/pre-teen suicides.

When you said AI (and in my view tech in general) are everywhere, I’m guessing this wasn’t what you meant…

My favorite explanation is that many new technologies end up redistributing wealth rather than creating it, which certainly tracks with both subjective and quantified growth in inequality on the same time period. However, a slightly more optimistic take is that tech is aligning production better with people's preferences, so that the same productivity enables people to live more distinct lifestyles that suit them.

[1] https://www.brookings.edu/articles/how-to-solve-the-puzzle-o...

If this is a good explanation, it begs the question of what AI might do to destroy productivity as well. If you’re constantly sexting with your AI girlfriend, who just happens to be extraordinary adept at tapping into your sexual proclivities, maybe you won’t get as many support tickets resolved as your boss was hoping.

More hypothetically, I would also expect that a world in which people spend a lot of time with screens strapped to their head, consuming an infinite stream of entertainment provided by generative AI, is not going to produce higher GDP.

Yeah, I think this dovetails with the idea that IT may be satisfying preference allocations without increasing overall production. Watching 10 movies a month on a streaming service adds much less to the GDP than going to the cinema 10x, but if the selection is better it might satisfy you more. Economists sometimes attempt to measure this with "utility adjustments" which recognize increasing quality in the same goods, but it's very hard for those adjustments to account for the hidden preferences of the consumers as opposed to objective qualities of a good or service.

Information goods like social media and streaming, financial services like pay-me-later, and conveniences like next-day delivery are all examples of activities that might suit preferences without showing up in GDP. They also may enable distraction, waste, reclusiveness and impulsiveness in ways we'd like to avoid as a society. At the same time they might also help some people feel more included and less lonely or trapped by circumstance.

The explanation sounds pretty familiar, so I might have already read/heard this from Klein, but would you mind sharing a link?

https://podcasts.apple.com/ca/podcast/the-ezra-klein-show/id...

No time off. No health care. Operate 24/7. No unions. No work safety concerns. No lawsuits over being unfairly fired. Control over exactly how something gets done or said.

If only the AIs will stop hallucinating or can consistently comply with policies …

I'm starting to wonder how much this matters.

People do crazy shit on the clock all the time. Company reps do not always adhere to policy 100% of the time either. People engage in office politics, coworkers accuse others of whatever, mistakes get made. LLMs happen to emulate all of this behavior.

In theory, we could replace everybody with AI and not much would be different. Productivity increase is debatable but cost savings would be immense. The question is how much insanity we're willing to tolerate as a result.

(...and seeing what fun ensues when there are more people than available jobs.)

Given our elected representatives, I don't think that's a problem. No offense to any party or person. But we've consistently proven that we can tolerate and welcome way more insanity than seems reasonable.

The second one is a lagging indicator. If lots of people bought their homes when they were cheap and are still alive, it will take time for the real impact to be visible.

Edit: Tracking real numbers in homelessness is also just extremely difficult.

It's also much harder to get servants in that house. I wonder why...

Have you looked for housing in Columbus, Ohio?

https://www.intel.com/content/www/us/en/newsroom/news/intel-...

In many places you can't be thrown in jail for being homeless anymore. Many cities have more housing and shelters and free kitchens than ever before. Some places even give you a smartphone and basic plan. Etc.

Measuring "Goods" in units or tons is bit simplistic. Almost everything is much better that it was 15-20 years back. TV, cars, phones, computers. This difference probably should be counted as 'extra', shouldn't it?

Driving, picking/packing, animal care, doing laundry - Absolutely nothing you mention is in any way some new 21st century job that didn't exist before. They're all just normal traditional jobs.

That person gave a list of tech they were talking about in their comment immediately afterwards: "speech to text, language translation, recommendation engines, relevancy ranking in search, computer vision, etc. and seems to be getting embedded in more and more processes by the day."

I'm not sure it's worth quibbling over whether we should use the term "everywhere" or "in many places"; the general point stands that it's found many different uses, and has done what effective tech does - fade into the background in many cases, just becoming part of our daily lives.

Sure, we're not seeing the off the wall predictions from the singularity crowd, but it seems to be tech that most people find broadly useful.

Translation is an interesting example: some labor has been displaced, but not nearly all because there's still value in having human eyes carefully checking the translation of high value documents. But free translation let's regular people translate things freely - a new capability which displaced no one.

However, productivity measures human productivity and human labor. The very cheap new translation modality is therefore completely missed by productivity measurements.

Meanwhile, there's /more/ jobs available right now, despite all of this. The US has hit a historic low in unemployment and wages are going up, leading to a decline in measured productivity. Productivity is output per dollar of wages, which means we twist our hands in anxiety when workers start doing better...

I lack the economics knowledge to do more than parrot the response I've heard to this, so take this with the appropriate level of "hmmm":

As I understand it, the counter-claim is that the measure of GDP mostly excludes exactly the set of things that grows absurdly fast.

For example, the measure of inflation may include the cost of a smartphone in the standard basket of goods, but not the fact the GPU of a smartphone (or Apple TV) of today, operating in double precision mode, can do more than the Numerical Wind Tunnel supercomputer in 1993 costing 100 million dollars.

Or that everyone has a free encyclopaedia a hundred times the size of the Encyclopædia Britannica.

And maps which for most users are as good as Ordnance Survey, but free and worldwide, when the actual OS prices for just the UK is… currently discounted to £2,818.17, from £4,025.97.

Or that getting you genome sequenced now costs a grand rather than 3 billion. Although that might not yet even be in the basket, I don't know where the actual baskets of goods get listed in most cases, and search results aren't helping — one result, on a government website, lists "health", but even digging into the spreadsheet didn't illuminate much detail there.

The UK basket of goods is here https://www.ons.gov.uk/economy/inflationandpriceindices/arti... and the various sublinks.

Maybe you design a wrench that is 1000x cheaper and faster to use and more reliable. Well, if it makes your car building operation 0.0001% faster, that's the impact. The details of the wrench and how impressive it is are irrelevant to any observer.

If having your genome sequenced leads to far longer or better lives then we would see the impact in productivity. Same with everything else on the list.

Source: This week’s Kurzgesagt, right?

My life is permeated by tech (and big part of it is AI) and made 100 times easier. I can buy a plane ticket to another country while waiting for a subway (did people use to take an hour to go to a special place, wait in line and talk to human to buy a ticket? I still remember this). I go there and quickly navigate in a city I know next to nothing about, find something niche, like cool local cafés in the area because GPS and google maps. I go to a restaurant and I can use google translate to understand the menu. I don't even need to type unfamiliar words, AI scans the image and translates it on the fly. The same google translate with speech recognition AI helps me to converse with a person when we don't share any common language. I can click couple of buttons and video-call my mum who lives on the other side of the world. If I need to buy something I need very rarely, I can order it online and not think where I find a shop that sells those things. Even if I don't know the right word, I can now ask chatgpt "what do you call in German that fancy thing you mount on the ceiling and attach lights to it?"

My life is _hugely_ more efficient thanks to tech and AI. Does it help me to contribute more to the abstract economic growth? I don't know, perhaps not. But I just don't care about GDP.

Government can print as much of the latter as they want; wealth goes up when we can collectively buy more and down when we can collectively buy less, regardless of how many dollarpounds that is.

But GDP is measured in money, and can only connect to wealth if we get get inflation right, but that's really hard because inflation depends on what you want to buy — childcare costs don't matter if you have no kids of that age.

That said, I trust the domain experts to get this right, even though the various governments may be incentivised to claim their own preferred numbers. Even at worst, they'll have thought of vastly more influences than I can even imagine.

AI is already so ubiquitous and useful that you blindly take it for granted without even thinking.

Spanish, for example, has many examples of words that are innocuous in one dialect and profane in another.

Bad translations are present in product names & descriptions of at least 70% of all products on Amazon and Ebay I've seen, and it doesn't look like it hurts the business in any way.

Big difference between "buy this $5 plastic crap widget despite its product description being barely coherent because there's only so many seconds you're willing to spare searching for 'Plastic Crap Widget' because Familiar Tech Company's algorithm puts CBSPOO ahead of VRIENGLU in this particuar search parameter combo" and "hello professionals in niche market. You will recognise the name of our product whenever you next have $xx,xxx to spend on these services, because we were the one whose ad was inadvertently extremely sexist"

Others, however, do care. I, for one, will pass on something that has bad translation because I take that as a proxy for the quality of the product overall.

Last year, for an important letter that had to be written in Japanese, a language that I don't know. Using Google Translate for that was unthinkable, because Google Translate is pretty poor and I had no way of checking and correcting the translated text.

If I was going to name the biggest good thing AI has done to humanity so far is the ability to read internet sites in other languages like Chinese (Google sucks at it, you have to use other tools, I use an app called "tap translate screen"). Also ability to do voice to text and translation at the same time on mobile devices (currently requires online connection).

As for the rest of your comment... please don't hijack other conversations for soapboxing on the industry as a whole. Instead, submit your post and open a real conversation.

I think the contention is with AI suddenly being redefined as only referring to language models, and the view that intelligence has been solved by these models.

There has clearly been a massive marketing push to label these models as the "one true AI", both from companies and from AI influencers. This is where the echo chamber exists, and it's easy to get stuck in it.

Maybe I am wrong and we have solved intelligence. But I seriously doubt it.

1. ChatGPT was released for general-purpose use. It's not a data science team at a FAANG company or healthcare or finance enterprise using ML for a specific business need. It's there for anyone to ask it anything.

2. A design decision was made to have ChatGPT output words in "real time" instead of all-at-once after a delay. To the user, that makes it look and feel like it's consciously and actively responding to you in a way that animated ellipses do not. I never knew what it would feel like talking to an AI, but when I first used ChatGPT, I thought: this must be it.

You know very well that that's not what GP is referring to. Speech to text, natural language translation, recommendation, and computer vision are all very useful things, but also were very much real and in consumer hands long before the current hype cycle.

Generative AIs are in their hype cycle. IMO the tech is overhyped to hell and back, but it will still probably yield better results than Crypto, there are legitimate uses for it. But those uses need to be ok with an 80% correct solution, which is not sufficient for all the things LLM hypelords are saying they can be used for and there is no path forward for closing that 20% gap.

Fortnite Bucks, Mt Gox trading cards, skulls of adult humans and guano also "have the potential to become the world's reserve currency", but it seems unreasonable to put the burden on proof on the people arguing that they won't be.

Fabulous, I'll be stealing that ...

AI only does bad things to me so far: surveillance, spam, fakes, search results poisoned, twitter and reddit closed up because of it etc.

Where is my automatic captcha solver? Where is a robot that will get me to a live person in a support call? Where is a spam filter that doesn't send all useful emails to spam? Where is a filter to hide fake reviews on Amazon? To fight against Amazon's crazy product ranking system? Such useful things are nowhere on the horizon.

> Where is a filter to hide fake reviews on Amazon?

While it[1] doesn’t hide, it can generate some insightful information about fake reviews for me.

Disclaimer: Not affiliated, only a happy person using FakeSpot since this year.

[1] https://www.fakespot.com/

Until then, you’ll get a lot of “it’s a really hard problem to solve!” Coupled with zero progress.

But what it already has begun to do, and will continue to do is change the way we interact with computers. The era of having a personal voice assistant that is capable, adaptable, and intuitive is VERY close and that is something that's exciting. Siri and Alexa are going to look downright primitive compared to what we'll have in the next 2-5 years and that is going to be VERY mainstream, and VERY useful for huge swaths of the population.

Crypto still hasn't proven itself to be useful in any way shape or form that isn't immediately over-shadowed by a different medium.

You’re treating it as a fact that LLM are going to replace existing products, at some unknown future date.

“In 5 years, all code will be written by AI”

“In 5 years, LLM will replace Siri and Alexa”

“In 5 years, AI will replace [sector of jobs]”

The thing that frustrates me about these statements is that you don’t know what AI technology is going to look like in 5 years, so stop treating it like a fact. It’s possible LLM are useful in all of these places, but we don’t know that yet.

That’s a fact.

I also know that voice-interfaces to date have been incredibly stiff and there is ample room for improvement. I know, for a fact, that having AI enable better voice interfaces will make computing better and more accessible. I have a hard time understanding how those are hype-driven comments and/or opinions.

We do know these things for a fact. Not being able to articulate exactly which breakthroughs will be most important doesn’t make it hype.

There doesn't seem to be a rush because it makes the implementation a lot more expensive, and those things are, I suspect, not profitable products (revenue sources) to their respective companies. They are a kind of enhancement to a layer of products and services; people take them for granted now and so you can't take them away.

A smarter Google Assistant would do nothing for Google's bottom line, and in fact it would cost more money to operate.

If it's not done right, it could ruin the experience. For instance, it cannot have worse latency on common queries than the old assistant.

All I did was hold it's hand, it wrote every line of code. You are living in fantasy land if you think we will be writing lines of code in 10 years.

I was with you until that sentence. No, LLMs will not write all our code and the reason is very simple: coding is easier than reviewing code. Not to mention the additional complexities and weirdness that we've always dealt with without even thinking about it.

We can see in Photoshop what's coming for developers: context-sensitive AI autocompletion and gap filling. Copilot but more mature and integrated, perhaps with additional checks that prevent some bugs being inserted. And troubleshooting, the area where I think we can profit the most.

but will ChatGPT help you debug and fix a production issue that came about due to a Kafka misconfiguration? will it be able to find the deadlock in your code that is causing requests to be dropped? will it suggest a path forward when you need to replace an obscure library that hasn't been updated in 5 years? will it be able to make sense of seemingly contradictory business requirements?

Wake me up when ChatGPT is able to write and maintain a POS system, or an online store with attached fulfillment management. Anything that goes beyond a fancy 100-line script. Anything that people actually hire teams of senior devs, business analysts and software architects for.

I like to think of capable LLMs as pf gifted interns. I can expect decent results if I explain well enough, but I need processes around them to make sure they are doing what they are told. In my industry thats enough to produce a noticeable productivity gain, and likely some reduction of employment as its a low margin cut throat business relying on low grade knowledge workers. I see the hype and honestly cant stand it, but its measureably impacting my industry and the world around me.

Except humans can transparently explain themselves and someone can be held to account when something goes wrong. Humans have the ability to have differing opinions and approaches to solve unseen problems.

An AI however cannot explain itself transparently and just resorts to regurgitating whatever output it has been trained on and black-box AI models have no clear method of any transparent reasoning meaning that it cannot be held to account.

Any unseen problem it encounters, it falls back to fixed guardrails and just repeats a variation or re-wording on what it already has said. Especially LLMs.

Personally, I think LLMs are a step forward, but I suspect that GTP-4 is close to the limit of what’s possible with LLMs. I don’t think we’re going to see AGI from the same approach.

Stone ages. That’s not 5 years from now. That’s today.

I can't trust GPT, and neither can you. But if it really can do all your coding for you, what stops your employer from replacing you with a secretary from a temp agency?

It's so stupid for engineers to say that ChatGPT codes for them. They are shooting themselves in the face. They are devaluing the entire profession. Why? My reaction to all those breathless online demos was to point out the difference between what they were showing and what an engineer really does. Your reaction is to act like being a prompt jockey is the new way of engineering. How does that give you pride in yourself?

I do and ChatGPT code is rarely useful for me. I can prompt it well enough to do language related stuff for me, but the code it can write for me is more like a highly custom boilerplate that I still need to refactor.

Even for green field private projects, at first it looks fine, bit the bugs are more likely to be traced back to these snippets than not.

Year of the voice assistant is getting close to year of Linux on desktop.

What you’re promising has been promised time and time again, received endless hype cycles then collapsed once people realised the limits of the technology. Yes, this time the tech is much more capable than what came before but I’m inclined to believe we’ll yet again find a limit that means we’re using it for some things but our lives still aren’t drastically changed.

Case in point, I asked Siri to change my work address. She stated that I needed to use the Contacts app to do that. This is not very helpful. The issue here is not Siri’s inability to understand what I want, it is that the Contacts app does not support this method of data input. Siri is also probably not very good at extracting structured address information from me via natural language, but the new LLMs can do this easily.

…which is something an LLM won’t help with.

“Just design an open ended API capable of doing absolutely anything someone might ask ChatGPT to do” is not the simple task you’re making it out to be!

There's a reason why people describe ChatGPT as a "research tool": you often need to do a bunch of iterations to get it to do the correct thing. And that's fine because it's non-destructive. But it's very far from a world where you can let it loose on a production, writable database and trust that it's going to do the correct thing.

No idea how well, so I assume "badly"; but the API is already there.

Intuitive to use? Or has intuition?

Google and Amazon have tried to sell theirs for a long time. And none were actually selling much. Amazon admitted to be selling theirs at a loss. Facebook has tried their own - and quickly cancelled them. Google's is in every Android device - and yet pretty much nobody uses them. Even Apple's Siri is more annoyance than help.

That something can be built doesn't mean it will sell or that people will actually want to use it. If you create a solution looking for an imaginary problem that your marketing thinks is what people want instead of a solution that solves a real existing problem, you do get a solution looking for a problem ...

Also, answering questions and communicating in natural language is the easy part of such assistant. For the thing to be useful it must be able to actually do something too. Which is incredibly difficult beyond the (closed) ecosystem of its vendor. Thirdparty integrations are usually driven by who pays the manufacturer for the SDK and partner contract (seen as a marketing opportunity), not by what the users actually want it to integrate with. Hoping for one of these with an open API that anyone could integrate whatever they want with, I am not holding my breath here.

OpenAI is already on it. The latest gen of GPT-3 and -4 are finetuned to respond to "do this thing" commands with JSON structured to:

- provide the name of a given function call

- provide arguments to that function call

it's "early stage", which in this case probably means "good enough to be useful within a month or two", given the rate at which these things have been developing.

Anecdotally, I've been playing with giving the models instructions like:

"When asked to perform a task that you need a tool to accomplish, you will call the tool according to its documentation by this format:

TOOL_NAME(*args)

Below you will find the documentation for your tools."

...and I've gotten it working pretty damn well (not even with the JSON-finetuned models, mind you). All you really need is python-style docstrings and a minimal parser and you're off to the races. I recommend anyone interested play with it a bit.

They couldn't have released this at a better time, I have about 30 plugins and i'd say it manages to get the right one about 90% of the time as opposed to about 70 with my hacked together version (but I guess I wrote it and know what to say so maybe that's a bit skewed)

@Tool

def add_todo(title, project=None) -> str:

    """
    Add a new TODO.

    Args:
        title (str): A brief description of the task
        project (str, optional): A project to add the todo to, if requested
    """

    logger.debug(f"Adding task: {title}")
    task = Task(tw, description=title, project=project)
    task.save()
    return f"Added [ {title} ] to the { project + ' ' if project else '' }TODO list."

Dream on.

This is not a technical problem, this is a business problem. Sadly a lot of engineers don't understand that.

The thing is, I've managed to get this working as an interface for a whole segment of stuff that was a pain in the ass before. My task list is all in one place for the first time, and it talks! With words! I have a pair programmer, who is excited to do stuff, on the command line, 24/7. They also have encyclopedic knowledge of anything that isn't a super deep cut, so I can move through more spaces and find solutions that I never would have dreamed of due to the cognitive load of sifting through textbooks and documentation just to create a [ insert more or less anything here].

If you're looking at the folks here who are getting excited and wondering "What's up with *them*?, this is it. It's not about the Next Big Thing so much as it's about "Holy shit, computers are magic again". For themselves.

Of course, I can speak for some of us. For sure, the hungry lets-make-a-startup folks exist and are currently working on doing that - and that's fine. But to me that's boring. Commerce and markets and economies are toxic to creativity. I've tried Bing-with-GPT and it's AWFUL compared to GPT-4, despite being sorta the same underlying thing.

I'm perfectly happy paying OpenAI to use the thing they built, for myself, for now. I am seriously looking forward to migrating to locally run models, once we get there (and we will).

“Intuitive to use” roughly means that it is easy for a human to interact with.

“Intuition” is the ability to understand something immediately, without the need for conscious reasoning.

I don't really see either of those things as a real possibility. Within my lifetime, anyway.

Seems like it has proven very useful for Stripe [0], Moneygram [1], TicketMaster [2], etc.

Unlike AI which continues to consume tons of resources to burn the entire world down to the ground without any viable efficient methods of training, inference or fine-tuning their AI models in the past decade with its chatbot hype and gimmickry [3], crypto does not need to consume tons of CO2 to operate, thanks to alternative and greener consensus algorithms available in production today. [4]

Being 'useful' is not an excuse to destroy the planet around untrustworthy AI models getting themselves confused over a single pixel or hallucinating in the middle of the road.

[0] https://stripe.com/gb/use-cases/crypto

[1] https://stellar.org/moneygram

[2] https://business.ticketmaster.com/business-solutions/nft-tok...

[3] https://gizmodo.com/chatgpt-ai-water-185000-gallons-training...

[4] https://consensys.net/blog/press-release/ethereum-blockchain...

Both have resulted in a bunch of hopefuls starting companies, in order to attract mountains of venture capital. Companies that will only have a loose connection to the tech that drives the hype.

Is there any insight from this observation?

Yes: be deeply skeptical of anyone claiming tech they are personally invested in is revolutionary.

For most people that's a promise for the future, but beside some translation tools (who are far from perfect) there is not much.

For instance: semantic search is and was a big topic, but so far even ChatGPT is not a real answer, Stable Diffusion is very nice if you want to produce some cartoon alike graphics or some porn deepfakes, but just not ready for simple common photo editing tasks. OCRs have gotten better, but still nothing that "by magic" makes a badly scanned piece of paper into an almost-native clean pdf and so one.

Yes, there is much progress and potential, but not much for real world usages.

Likewise, speech to text and language translation are more available, but they're still pretty bad. And computer vision is much better than 10 years ago, but I wouldn't bet anyone's life on it.

And yet, the hype train is still gaining momentum. Having been through more than my share of AI winters, I can feel another one coming and it's going to be bad.

I've predicted 9 of the last 1 AI winters.

As always, sounds cool. Actually do some of it and then let's talk more.

No wonder that its adoption for legal purposes didn't go anywhere.

Some of the applications crypto folks were going on about: decentralization (of course) until the IRS started taxing your crypto transactions; helping 3rd world countries with weak currencies (partially happened); international trade (quickly became untrue and monitored by government agencies), ledgers for tech companies (they can all already build audit trails. i've yet to see many applications where companies are willing to give up control over their trust to a 3rd party system with no scrubbing functionality. looking at you, AWS status page).

Like I said, same vibe, different applications. Until some are built it's all just hype and conjecture. The applications we've seen work well are already well-accepted for their faults (content generation, summarization, etc)

Want to elaborate? Bitcoins monetary policy is beautiful in its simplicity and predictability. In that sense, it stands in stark contrast to fiat.

Real interest rates are typically positive, so no, you are not.

If you are a machine-learning practitioner, you should be familiar with all of those techniques and how they are used so that you can solve practical problems with them. But if you just read about AI in the news and figure you're going to found the next great startup and make a billion off it, you'll probably start by feeding a whole bunch of data into Tensorflow and then getting useless garbage out of it.

This hype bubble is specifically about LLMs, extremely large-parameter transformers that are trained on all the data OpenAI or Google can get their hands on. And then supposedly if you ask them the right questions, you will get useful answers back. For people that put in the time and experimentation to actually find the right questions and the right applications, that will probably be true - but the hype is that this will change everything, and it most certainly will not, just in the same way that beam search is frequently useful but it definitely does not change everything.

But slick promoters will nevertheless manage to use people's lack of knowledge to redirect billions of dollars in capital into their and their employees' pockets, the same way that slick promoters used crypto to redirect billions of dollars in capital into their and their employees' pockets.

OpenAI which is funded by Microsoft and promoted by Microsoft account executives creates hype as if it's open although nothing, including its so-called open-source Whisper, is open. People feeding Microsoft pretend that "they" are revolutionizing the world. NVIDIA and Microsoft are making money out of these large models and positioning the bigger as the better.

In the case of AI: both potentially quite useful (unlike crypto) and incredibly, toxically overhyped (just like crypto).

Ironically, the fact that a lot of people think "You it's actually kind of useful sometimes, therefore you can't compare it crypto/web3" is part of the engine that drives the hype.

The tech religious overhype train is here to stay. There has never been a more established need for calm honesty.

Do you understand what a massive impact that has been? It has disrupted one of the largest industries on the planet, which is drug trafficking.

People are buying drugs on the darknet, who would never buy it on the streets or want to be associated with regular drug users.

[1] https://www.unodc.org/res/wdr2021/field/WDR21_Booklet_2.pdf

"Cocaine News" with Donald Trump Jr. | The Daily Show:

https://www.youtube.com/watch?v=47yFRXZqB0g

Don Jr. Swears He’s Not on Coke—He’s Just ‘Impassioned’:

https://www.thedailybeast.com/donald-trump-jr-is-tired-of-co...

So speaking of echo chambers…

Overall economic productivity didn't shoot up in the last decade despite the dramatic progress we had in software and hardware (e.g. [1]), and it's not clear that AI/ML will dramatically change that. Yes searching pictures on my iphone by text is convenient and Netflix recommendations might be more addictive, but the path from that to ubiquitous economic prosperity, safety, and comfort (the technoutopia many here are striving for) is not clear at all. It's also not very clear if those marginal improvements are worth the substantial share of total human brainpower thrown at them.

[1] https://www.aei.org/economics/good-news-bad-news-on-us-produ...

    AI is just like crypto, but better!

Both don't seem to work.

no

calling something "hype" should not be a stand-in for data

Consider how power structures (eg nation states) may change in such a future.

Well, apart from the fact that chatGPT is really incapable of developing a thought, and also apart from the fact that half will fail to delete sentences like "I'm a language model, so I can't..." (insert gist of question here), it's painfully obvious if something is LLM generated.

The moment a sentence like "it's crucial to remember" pops up, I know what this is. Then, there's also the element that it always sounds like it's speaking to a child, and it avoids actually saying things unequivocally without some sort of disclaimer, as the legal department's CYA filter will ensure.

I remain thoroughly unimpressed by the entire venture. If this is Skynet 1.0, we're all safe for centuries to come.

Students who pay the $20 a month for it and are aware of its limitations will absolutely use it and it won't be obvious.

I just asked Chat GPT 4 to explain the religious significance of the Wizard of Oz as a literary critic. Here's some of what it gave me, it doesn't write anything like you claim it does:

"Moreover, Dorothy's companions -- the Scarecrow seeking a brain (wisdom), the Tin Man seeking a heart (love/compassion), and the Lion seeking courage (strength) -- symbolize spiritual virtues that are often extolled in religious texts. They embark on this quest together, mirroring the communal aspect of many religions.

The slippers (silver in the book, ruby in the film) can be viewed as sacred objects, or relics, that assist her in her journey, providing divine protection and eventually leading her to salvation (returning home).

Finally, the revelation that the Wizard is a mere mortal, and that Dorothy had the power to return home all along, imparts a spiritual lesson often found in religious narratives: the divine or the sacred is not external, but within us."

If I was a student I could have easily expanded on these concepts (with or without GPT) and turned in a good essay.

This isn't a field like engineering where there are objective right and wrong answers and anyone dies if you pass the students who are not so great at writing essays on literature.

Depending on the assignment you aren't necessarily expected to read anyone else's take on a book and you aren't expected to make sure you are saying something that hasn't been said before or anything like that.

You are simply expected to analyze the book and offer an interpretation.

And it's not like that's the only way to use the AI. With a few minutes of effort, I just got CHATGPT to write an essay using "post-colonial theory" to interpret the Wizard of Oz, which was pretty interesting.

Also big chance to get a C or a D when you came up with a "novel" approach that the shitty and boring book is actually shitty and boring.

Hell no school was there to stop making your own interpetations different than the official one.

Damn, even on the fucking retarded drawing lesseons (where probably half of stuff was drawn by parents) the teachers would deduct points for any individual style.

I was thinking of getting an MBA, but does it even get any better for "adults"? Arent you just tought to repeat some schematics, which often are bullshit.

"As a literary critic, describe how Dorothy in the Wizard of Oz is a religious figure."

The divine being contained within I would think would match Buddhism pretty well.

The reference to relics is too vague to pin down to any religion, there's probably lots of examples of it in lots of religions. If I had to defend it off the top if my head I'd compare the Ruby slippers to the "holy moly" herb Athena gives Odysseus to defend him from Circe.

If anything I think GPT went wrong saying strength is one of the virtues associated with the Lion. It would be much easier to focus on courage and say he needs to learn to be like a brave apostle who says things like "Yea, though I walk through the valley of the shadow of death, I will fear no evil; for thou art with me:"

My point wasn't that this essay was particularly good, necessarily, only that it was was good enough for undergraduate work.

Michael Berubé has this story where he says, he once came early to class and overheard the students make great arguments about movies and shows they saw last night, discussing them heatedly. Then, when the lesson started, all the arguments turned bland, banal, reproductive.

Obvious conclusion: They -can- very well produce good insight, but the college and school systems discourage it. They reward them for repeating ideas they read in books about their things, or what the teacher said; an original idea is dangerous, because they're responsible for it themselves, and if the teacher doesn't like it, they'll get punished for it. Safer to say, "Miller said..." and shove off accountability to someone published.

I can see why you believe identifying GPT-generated text is easy. This is because techniques like prompt engineering, few-shot learning, and fine-tuning aren't known and used extensively yet. For instance, with a 32k model, you could input all your previous writings and the instruct gpt to mimic your style—even down to the grammar mistakes.

This requires having a massive amount of previous writings to input, otherwise gpt struggles actually differentiating styles enough to generate it consistently in the same way a human would. Most students do not have enough personal writing data to train from.

This also excludes other strategies, such as getting every student to write on paper in a supervised environment and using this to guide your pattern assessments of submitted electronic works. It's very difficult for people to remain consistent and implement their own style in gpt generation. You can ask the many creative writers who are trying to use gpt for stories, and how many of them have to treat generation as an extreme rough draft of plot points at best.

I think absolutely anyone claiming that detecting LLM generated text is easy is flat out lying to themselves, or has only spent a few tokens and very little time playing with it.

Take semi-decent output, give it a single proof read and a few edits... and I don't fucking believe anyone who says they'll detect it. They absolutely will detect some of the most egregious examples of it, but assuming that's all of it is near willfully naive at this point.

They aren't going to risk getting expelled. Schools have done a good job of putting the fear of God into kids to not use chatGPT. Better to just not turn a paper in than to be accused of plagiarism.

All chatGPT shows to me is we have a ton of smart, incredibly closed minded people that know what they know and they think they have it all figured out.

My paper would be easy to spot if chatGPT helped because the writing would be so much better. The thoughts would be much better organized.

> "I'm a language model, so I can't..."

You won't catch the clever students who programmically remove these (e.g. using Langchain).

They gonna ruin even this technology with their hype marketing bullshit.

The issue I have with all the hype is not the technology itself, it will stay in one form or the other as a better interface for generalized instruction communications, but rather the scams and frauds that come with it.

Empty marketing promises where everybody advanced enough realizes that it cannot be true, and as a concept GPT is just throwing more averaged neurons to the problem instead of training more specialized expert transformers for multiple knowledge categories. Anybody remember IBM watson?

Why I always say that I don't do AI work is because people tend to think that sensorics (and neural nets that reduce a min/max problem space) already is AI. And I think it isn't. AI is where the bayesian approach is the bare minimum to deal with strategical decision making processes.

(Which probably makes this comment go to hell with downvotes but who cares :D)

I’ve already used the openai api to automate several genuinely difficult things for myself. Mostly acting as a translator from natural language to structured output.

I do agree it is massively overhyped and there will be an inevitable sentiment correction.

Can I ask what exactly that means/does?

With the Wolfram plug-in, this works and provides good data! It stops three short of the goal probably due to rate limiting, but I think you can get the point: https://chat.openai.com/share/9d6695a9-5ba8-44d8-9ec8-11fcba...

This same kind of query works with any reasonable structured data format.

https://www.wolframalpha.com/input?i=Give+me+a+JSON+document...

My team raised a support issue with one of our suppliers due to some unexpected API behaviour and got an unusually flowery reply that completely contradicted the API documentation... fairly sure that was ChatGPT.

Honestly not that bothered about LLMs as they could be helpful in customer support particularly when agents might not be fluent English speakers (or just help when you're trying to be polite in adverse circumstances), but some basic proofreading would help. And don't let it hallucinate APIs.

Crypto was a great promise when it was invented and one can make the argument that it could have so many real world uses but it failed to live upto that expectation and one of the reasons is that it was over hyped way too quickly and ultimately became a tool for get rich quick, speculation, scams, dark web payments etc.

AI is already much better with its use BUT the hype is dangerous and we need to be careful. I see a lot of people starting "X-GPT.com" apps and touting 10K MRR in 2 months and what not. THis is what worries me. Every Tom, Dick and Harry is starting yet another AI tool. It can't be because they are so excited. It is because they see it as the new Crypto to get rich quick.

Overall, I think that AI is the new Crypto unfortunately not because it has no real world application (it does and is lot better than crypto) but because of the hype and everyone trying to cash in on it.

I’ve also recently been involved in designing a DevOps /Docker deployment pipeline for a customer. They use Java and I haven’t used Java in decades.

Before I would have just done my POC using a Python or Node container and rely on the fact that they knew Java well enough to get the concepts. But I used Java and started the chain of questions “answer all question based on talking to someone who doesn’t know Java. Explain everything step by step.”

In both cases, ChatGPT will usually get me 99% there. But I have to keep trying things and giving it the error messages and iterating.

Of course there is the hallucination issue.

On the other hand, I’ve done a lot of work professionally with old school chatbots integrated with web pages and call centers where the only intelligent component is that we could parse out parts of speech (nouns, verbs, adjectives, etc) and only search on those.

I would never recommend putting an LLM style chatbot in front of a customer. When I work with customers - especially in the government - the questions and answers are heavily vetted before being put in production.

They would never take a chance that either the customer could jailbreak the chatbot and have it say something and trigger a political argument about “bias” or that it would give incorrect information about a government benefit.

It seems to give answers that are 100% incorrect and when told so says "of course you're right, here is the right answer". The only stuff I'd use it for is if I already know exactly how to write the script and I'm just using it to type it for me quickly because I don't remember if aws_instance is the correct spelling or aws-instance in terraform...

But with code, it’s easy enough to prove correctness just by running it.

That being said, the one bug I find consistently with ChatGPT is that with the AWS APIs, all list type methods pagínate and you have to account for that. Python/boto3 have built in paginators and it doesn’t.

This an insidious bug because things will work correctly in a dev account with only few resources. But will fail in hard to debug ways in production.

What’s even worse is that ChatGPT “knows” the pattern and will correct itself once you say something like

“This won’t work with more than 50 roles/ec2 instances, etc”

I may not know always know the correct API or CloudFormation/CDK/Terraform syntax. But if it gets it wrong, I can read the docs and correct it.

Providing sources doesn’t usually help. ChatGPT consistently makes up sources.

Certainly with code the proof is in the pudding, but most recently my problem was "I need to create AWS monitors in Datadog to alert when a region is down." ChatGPT was hopeless but bard was able to point me to the exact doc explaining how to set it up.

You realize it’s still here, adoption is increasing, utility is increasing, etc

Just because it’s not a hype cycle does not mean it’s dead or even close to dead.

Genuinely curious. Where ? Remember it's been 15 years already. Few anecdotal examples are not good enough.

This stuff can already do impressive things and its only getting better.

Douglas Hofstadter and Geoffrey Hinton both think that we are on the path to humans eventually being surpassed.

I would urge everyone to hold back their instinctive reaction to the usual SV hype and go and try GPT-4,Claude+, Mid Journey, RunwayML for a few weeks and come to their own conclusions.

And the answers below, reducing the industry to « drugs » go in the same vein. Stuff like Chainlink working with Swift is not common knowledge and even if it is, it is considered as another nothingburger.

A 10% efficiency boost that some programmers are experiencing could translate into an extra 5 weeks off if you are smart about it, so it is quite life changing for some.

Would you mind expanding on that a bit? I've largely had great experiences getting what I want out of ChatGPT. But I've been continually surprised by the number (and variety) of people who don't see the utility of it.

For the non-chat interfaces, I imagine a whiteboarding session with a really competent intern at the board, rapid prototyping / wireframing that you can play with "live" and refine far further than you could IRL, but still ultimately prototyping.

> I've been continually surprised by the number (and variety) of people who don't see the utility of it.

If you _don't_ do it this way, you can easily fall into all sorts of time wasting anti-patterns; if you try to trick it, or allow yourself to be easily fooled by it, get stubborn & closed minded, pedantic and argumentative or whatever, well, there are lots of examples of how those sorts of interactions go in the training data too, and it will just as happily go down them as any other.

I don't even start talking to it without the first instruction being "answer all following questions in the shortest form possible"

This cuts out 90% of the useless output such models generate.

I wonder if this is a regional thing, because traveling between the US Northeast and West coast I've found entirely the opposite.

I've had non-technical friends reach out to me in a panic worried that AI will disrupt humanity in just a few years, and even my 90-year non-technical grandmother recently remarked about her fears about what AI would bring in the next 5 years.

And among technical people: ever since I posted getting an AI related role my linked in I've been bombarded with old acquaintances trying to get ahead of the AI boom.

The funny thing is that I personally think that "AI" is useful but wildly over hyped right now. I do think it has some uses, but they aren't going to change the world in any fundamental way (but hey, if I'm wrong at least I'm in the right field).

Meanwhile, there are tons of applications you use everyday (and have for YEARS) using “AI”/ML for document search, text suggestion, NLP/NLU, intent recognition, STT/TTS, image similarity/classification/search and a myriad of other tasks. LLMs have sucked all of the oxygen out of the room and there are tons of “AI” companies/“engineers” now who have never even heard of any of these and are doing all kinds of bizarre (wrong) things to wedge these tasks into LLMs.

I cringe when I see people all of a sudden jumping on the “AI” hype train thinking an LLM (or even the ML approaches I listed) is a universal solution to everything. They are interesting and have use cases but please stop.

Fair.

I'm definitely big on where it will be in the future (Iain M. Banks quote about Minds being the next thing to gods and on the other side), but there are a lot of grifters who are easy to spot with the following thought experiment:

If ChatGPT could actually, to use an example I've seen, "write a best selling novel", why is OpenAI selling you access to the API instead of writing all those books and selling them directly?

The grifters say things like "buy my guide to learn how to use ChatGPT to write a book for you", overselling the capabilities of ChatGPT by a large margin.

Anytime someone says "buy my guide to becoming rich", that should set off warning signs. I've only heard it being true once ever, but even that might just be a case of a random dice roll we wouldn't have heard about if it had lost: https://en.wikipedia.org/wiki/The_Manual

That said, one obvious difference between OpenAI and Intel is that OpenAI has full control of both the model and all the hardware the model is running on.

By the time someone has read a second paragraph, they have internalized what was at the beginning, and to be told that those two paragraphs were fiction is now to attack the reader instead of the text.

As though reading were so laborious, there's a sunk cost fallacy.

One thing I will say is that it is a decent editor. If you feed it a document, it will produce pretty good suggestions about improvements etc.

It's supposed to be the other way around.

Wikipedia on Hinton:

>Hinton received the 2018 Turing Award (often referred to as the "Nobel Prize of Computing"), together with Yoshua Bengio and Yann LeCun, for their work on deep learning. They are sometimes referred to as the "Godfathers of AI" and "Godfathers of Deep Learning"

Is there anyone at the Turing Award for AI level who's not impressed I wonder?

The story of 2023 is the CEO or another C-level boss rushing to their ML team and excitedly telling them to scrap all their plans because they need to integrate ChatGPT AI "yesterday", while the ML team roll their eyes and laugh behind his back.

I'm sure it happened to you too.

We're still waiting ...

Besides the science behind it, currently it feels like the same hype as crypto a couple years before.

A week after ChatGPT was out and plenty of people were already using it for writing code, emails, and plenty of other tasks. It would be weird to argue that AI is not going to have a massive impact at many levels.

Now there are plenty of challenges to overcome of course, but I have no doubts that something that useful on day one is going to have a big impact once we really understand how to integrate it to various products.

There WILL be job losses but it's not going to be the kind of people who hang out on HN. I can't think of any reason why you wouldn't have an AI taking orders at the drive through, handling customer service calls / tech support, etc. Any job that consists mostly of having the same simple, repetitive conversations is going to eventually be cheaper to have a computer do.