Fearing ChatGPT, Google enlists founders Brin and Page in AI fight

Threatened by upstart OpenAI, Google pledges to fast-track new AI projects.

See full article...

 

[xoe]

I also suspect that ChatGPT is about to hit a wall, much like autonomous driving. Larger datasets will yield incremental gains, but replacing an educated human on novel thinking problems is going to be a massive hurdle.

So, the way I understand it, Technology Life Cycles tend to follow an S-Curve and sometimes a repeating S-Curve. When this curve is followed the pattern is like this: when a new technology is developed or invented, there is a period of exponential growth in our knowledge of and ability to exploit the technology, this is followed by linear growth, then a tapering off of growth. The thing is, while observing the growth as it happens, it is only really possible to determine which phase we are in but determining how close we are to the next phase is incredibly difficult.

Large Language Models will "hit a wall" sometime after the current phase (as best I can tell, I have made no effort to do a quantified analysis of the growth of LLMs) of exponential growth ends, but when that will be is anyone's guess.

 

[kaibelf]

Hurry up Google, and rev up the spaghetti throwing machine! The wall will be delivered by Friday.

 

[sigmasirrus]

The availability of better GPUs and tensor processors (and a tsunami of funding) has allowed a very rapid increase in the number of variables (parameters), growing with a doubling time that is now four to six months. Prior to about 2018 the doubling time followed Moore’s law-about eighteen months. Once the parameterization is complete the training to a specific task goes quickly. This log-linear graph shows how the slope is now almost vertical.
View attachment 50685

For reference, I think the brain has some 86 billion neurons. And DuckDuckGo tells me that we peak at about 15,000 synapses per neuron when we’re toddlers (those get pruned as we develop our senses and language, and other formational learning). If you want to round that off to 100 billion neurons you’d get 1,500 trillion peak synapses. If each synapse could be reduced to one of these parameters in these artificial neural networks, that would at least give us some idea of where we’re at with raw compute power. There’s probably a lot of simplifications there. Simplifying things to 1:1 synapse to NN parameter ignores neurotransmitters, genetic activity, cellular activity within the neuron, and probably some unknown unknowns. Not to mention, just because you have enough parameters for all the synapses, doesn’t mean you know how to connect them. But it’s still interesting that it seems like maybe we’re getting a little closer to the neural processing of the brain?

 

[iamai]

I wasn't impressed with the technology until I asked ChatGPT to write a sonnet about a parrot that has amnesia. The result was not very good. But it did it. It wrote a f*cking sonnet. Iambic pentameter, proper rhyme scheme, even some word play and humor. And it's been around only for a few weeks. We're doomed.

 

[Bongle]

Google has self driving cars. I'm very confused by this comment.

Waymo, deployed in a very limited fashion in SF and a suburb of Phoenix, is nowhere near the hype that was being thrown about in the mid-2010s about AVs.

ImageNet and deep learning had blown everyone's minds in 2012, and everyone and their uncle was investing in AV cars since even university clubs could put together convincing demos, and then it turned out that final couple percent of functionality was going to be insanely difficult.

It's entirely possible that getting ChatGPT to "correct enough often enough to be worth paying for" instead of "very impressive and very confidently wrong, but very fun to play with for a few minutes" will be the same story as getting AVs from "will go 2,000 miles between interventions" to "highly reliable transportation system".

 

[inputbreakdo]

We’ll know we’re safe when SkyNet gets listed on killedbygoogle.com.

 

[Kamus]

Please tell me you don't think you could run a company the scale of Google off an AI model that fits on a cheap flash drive?

No Neo. I'm saying that when these models are ready... We won't have to.

 

[Deleted member 330960]

It's not that Google is so far behind on AI development IMO like they're not doing anything. It's because they put restrictors on. Google does have Lamda so they have people who understand what to do. It is more Google has restricted teams on what they should use their AI for and how to train it. It is the legal implications. Right now Microsoft is investing billions into ChatGPT instead of making their own team maybe for the reason if there's some massively immoral action happening they can just wash their hands clean. Where as Google can't wash anything clean because they have too many services that are publicly facing that can feed a said AI. So if the public has sudden backlash against AI doing XYZ. They're stuck in that sinking ship.

Do you think there are copyright issues in the legal implications?

I fear that chatgpt will be the end of the open web because any document that you have or any knowledge that you publish will be better when consumed using a chatgpt type system. I am a complete AI and legal novice, but I think that the protection copyright offered when making your work public is basically gone (in many cases already, in others maybe in the future).

 

[Voix des Airs]

No Neo. I'm saying that when these models are ready... We won't have to.

This is just wrong.

In an earlier (now hidden) post you said:

These AI models fit in a 2 GB file, and soon, maybe even a 200MB file. We won't need a centralized service for most of our queries.

According to Wikipedia, GPT-3 uses 175 billion parameters requiring on the order of 1TB of storage (so I presume doubles). So right there you are off by a factor of 500. And GTP-4 is targeted at around 1 trillion parameters. The only reason that GPT-3 is better than GPT-2 is because it's bigger. And there is widespread skepticism (including within OpenAI) that GPT-4 will be much better than GPT-3. It might be but everyone expects these things to hit a wall sooner or later.

But even without these concrete figures you are wrong. There are fundamental theoretical limits on how much information can be stored in a given amount of storage. I was going to try to explain it here but in searching just now for something to cite to this popped up. So just watch that.

 

[barktrees]

Waymo, deployed in a very limited fashion in SF and a suburb of Phoenix, is nowhere near the hype that was being thrown about in the mid-2010s about AVs.

ImageNet and deep learning had blown everyone's minds in 2012, and everyone and their uncle was investing in AV cars since even university clubs could put together convincing demos, and then it turned out that final couple percent of functionality was going to be insanely difficult.

It's entirely possible that getting ChatGPT to "correct enough often enough to be worth paying for" instead of "very impressive and very confidently wrong, but very fun to play with for a few minutes" will be the same story as getting AVs from "will go 2,000 miles between interventions" to "highly reliable transportation system".

This is like someone saying in 2005 that "pocket computers" were a bad investment because they've been around for 15 years but have failed to catch on.

That last 2% is really hard, but Waymo and Cruise have whittled it down pretty hard. Maybe we're down to the last 1.4% now. It's going to be here sooner than you think, and it's just as big a deal as everyone thought it would be a decade ago. It's just taking longer to crack than we thought.

 

[Kamus]

According to Wikipedia, GPT-3 uses 175 billion parameters requiring on the order of 1TB of storage (so I presume doubles). So right there you are off by a factor of 500. And GTP-4 is targeted at around 1 trillion parameters. The only reason that GPT-3 is better than GPT-2 is because it's bigger. And there is widespread skepticism (including within OpenAI) that GPT-4 will be much better than GPT-3. It might be but everyone expects these things to hit a wall sooner or later.

But even without these concrete figures you are wrong. There are fundamental theoretical limits on how much information can be stored in a given amount of storage. I was going to try to explain it here but in searching just now for something to cite to this popped up. So just watch that.

I meant AI models. I'm not talking about ChatGPT.
Stable diffusion was trained on 100 Terabytes of images, but the model itself is only a 2GB file, and I've been running it locally for months now.

Here's an interview with emad, the CEO of Stability AI. I think you'll find it insightful:

View: https://youtu.be/YaRdZ7d6o2Y

 

[Stochastic]

The skeptic in me wonders if this is just another "self-driving car" moment.

Remember when that was just around the corner and threatened to disrupt the entire transportation industry within a few years?

It also kind of reminds me of Google's "code red" drive for everything social.

So the most likely outcomes are this is a nothing-burger or Google will fail to execute.

Although I somewhat agree with your larger point, I want to note that self-driving cars are no longer theoretical.

 

[FerociousLabRetriever]

I don't think I made myself clear. I'm not saying that ChatGPT is ready to replace those people now. This is just a very small taste of what's to come.

But your prediction of 15 to 20 years is far too pessimistic:

In March 2016, the computer program AlphaGo, developed by Google DeepMind, defeated Lee Sedol, one of the world's top players of the board game Go, in a best-of-five match. At the time, this was considered a major milestone in the field of AI as Go is considered to be a much more complex game than chess, and many experts had believed it would be many years before a computer could defeat a top human player.

After the match with Lee Sedol, the team at DeepMind continued to work on improving the AlphaGo program. In October 2017, they released a new version called AlphaGo Zero. This version was able to achieve superhuman performance by training itself, rather than being provided with a dataset of human expert games, which was the case for AlphaGo. In just three days of self-play, AlphaGo Zero was able to reach a level of play that was on par with the original AlphaGo, which had trained for several months on expert human games. In 100 games against the previous version, AlphaGo Zero won 100 games.

To give you an idea of just how pessimistic your prediction is, AI experts used to say that AGI was 100 years away, decades ago, last decade? 30 years away. These days? They say 2030. (I say 2029, just like Ray has since the 90's)

ChatGPT already feels superhuman in some ways, the next version, will probably be better than us at almost anything "thinking" related.

It's always amusing to me, that the first thing most people do when they see a breakthrough of this magnitude is down play it's significance, and point out things like: "Well, but it's nowhere near a human at X, and Y"

This happened with Deep blue, Watson, Alpha Go, and now ChatGPT. It was bad enough when people did that in the past, but now? We're literally about to be surpassed forever in intelligence by machines, and people are still downplaying its significance.

It's all a matter of perspective. To you, expecting an exponential rise in this technology to the point it replaces professionals very soon is the baseline case, so anything short of that is pessimism. To me, the baseline case is this technology becomes a useful tool to professionals to make them more productive, and that real value is still years out with much needed refinement. So, relatively, your expectations, which are in line with popular press articles, is sensational and optimistic. I mean, you called it 'superhuman' (not true) and predict it's next version will beat humans at thinking.

You've made a comparison to AlphaGo to justify your optimism, but you're comparing apples and oranges. Computers have been beating humans at chess for decades, it's no surprise to me that models have evolved to tackle ever more complicated games, including being able to develop strategy from scratch (rather than say the traditional approach of massive databases). A really exciting development, no doubt. Still, that is not even in the same stratosphere as replacing doctors, teachers, accountants, lawyers, and engineers when facing novel problems in complicated and dynamic contexts.

To give you an idea of just how optimistic you are, people were predicting automated driving was going to be widespread years ago and we're still nowhere close to that happening. Not only are truck drivers not on the way out, their salaries are on the way up, and there is an acute shortage of them. It turns out replacing humans is much harder than most people think. Human truck drivers may eventually be out of jobs, but it turns out the time frame is MUCH longer than many people expected.

What's amazing to me is how people think every new technology is going to revolutionize the world overnight and continue to think that way after most turn into incremental improvements ;-) The examples you provide support my point. Deep Blue, Watson, AlphaGo all mark incredible advancements in technology, and yet their underlying technologies didn't replace humans, they provided us some better tools to be more productive. They didn't even radically change most professions. None of what you listed are even close to the scale of change you are so confidently arguing is just around the corner. At least point to more significant milestones such as the advent of computers or the internet, and yet, even these took decades to make transformative change, and even still, not on the scale you're arguing.

I don't think you appreciate the significant gulf that ChatGPT needs to cross before replacing real professionals, especially when it comes to dealing with edge cases or dynamic situations (e.g., when the pandemic hit) for advanced professions. I don't think you appreciate the significant legal or regulatory hurdles that will soon be in its way.

Don't get me wrong, as a tech enthusiast, I find these developments fascinating and believe it'll be a very useful tool. I just don't see the kind of revolutionary change on the timeline you're expecting. The deepest irony is I suspect ChatGPT would actually provide more compelling and nuanced arguments than you are providing here.

 

[Voix des Airs]

I meant AI models. I'm not talking about ChatGPT. Stable diffusion was trained on 100 Terabytes images, but the model itself is only a 2GB file, and I've been running it locally for months now.

And that replaces big data and the likes of Google search how?

 

[Smashing Young Man]

Watch Google start floundering around like a chicken with its head cut off like Meta did when faced with more innovative competition.

 

[xcodemustdie]

I was under the impression, due to sites like Arstechnica that Google was one of the leaders of AI. So I doubt they are too far behind the curve.
What is worrying though is this will no doubt create mentalityfor companies like google,facebook,Microsoft,apple and 100 startups to lets just rush headlong into this all guns blazing without caution in search of the next trillion dollar idea.
Is this being regulated? I doubt it
If there was trillions to be made in researching nuclear bombs governments would have some restrictions I assume, yet this which is just as dangerous (maybe even more dangerous because the path to creation is prolly lower) theres bugger all restrictions.

 

[hambone]

Notably, Google searches return not "answers" but "results." This is inherently comparative. The human searcher is invited to compare and assess several sources of information and make some kind of informed choice about what to rely on. This is a check and balance on bad information, albeit with flaws. While many (most?) Google searchers consider the range of results, it's also true that no amount of information can save a dumb person from themselves.

ChatGPT, by contrast, seems to want to spit out not "results" but "answers." This seems far more dangerous because it leaves the critical thinking and comparison of information sources not to the human but the AI, working invisibly and making choices and selections in ways the human can't see or assess. That's a wholly dependent relationship on the AI getting it right. And to many people it looks right and sounds convincing! But as is obvious to any subject expert or professional, ChatGPT "answers" rely on a lot of half-truths, simplifications, gross reductions, over statements, a lack of context, and weird fragmented thoughts. It's a bit like saying "yeah I've read that book!" when really all you did was read a review of it.

In this regard, I can only see ChatGPT being useful (and socially responsible!) search tool if any "answer" it provides also, at a minimum, comes with a comparative statement of the variety of issues and approaches that are related to the question being asked, and prompting the person to explore these to form a more rounded opinion. In other words ChatGPT can't leave people with the impression of receiving facts so much as contingent summaries.

 

[runemail]

Google's gotten too big and unfortunately... evil.

A punch in it's 149 billion per year empire is good and about time.

Sundar Pichai have been a disaster for the Google we used to love. 100% uninspiring. It's good news if Brin and Page take more control.

 

[sac_cw]

ChatGPT. For every company who thinks their ISP's India call center provides excellent and responsive answers to their technical questions.

 

[DragonDeborn]

Existential threat? Google has a whole suite of products to cushion them while they adapt and catch up to ChatGPT and its ilk. The Big G ain't going anywhere.

Any chance a free cloud storage provider with reasonable customer service and policies while threaten Google for a bit, though? With a little competition, Google will quickly realize people don't take kindly to storage services cutting them off from their data after advertising themselves as safe places to store said data.

 

[MihaP]

Existential threat? Google has a whole suite of products to cushion them while they adapt and catch up to ChatGPT and its ilk. The Big G ain't going anywhere.

Any chance a free cloud storage provider with reasonable customer service and policies while threaten Google for a bit, though? With a little competition, Google will quickly realize people don't take kindly to storage services cutting them off from their data after advertising themselves as safe places to store said data.

Thing is, once someone develops a functional "personal assistant" AI that's a step (or couple) above ChatGPT, everyone will want one.

Microsoft wants to be one. Google wants to be the one. Chinese equivalents already are trying out a few. I'm pretty sure we'll soon see a similar announcement from Tim Cook.

And when one succeeds, the rest will get left behind in the dust.

 

[davidlars]

I don't think I made myself clear. I'm not saying that ChatGPT is ready to replace those people now. This is just a very small taste of what's to come.
.....
ChatGPT already feels superhuman in some ways, the next version, will probably be better than us at almost anything "thinking" related.

ChatGPT is not thinking, though. I think this statement explains many of the downvotes. I think most agree that we see a lot of impressive and eye opening output from these models and products, but the kind of AI we see today are still narrow in their capabilities and can't be used for completely different kinds of tasks. ChatGPT can't, for example, play Go or drive a car.

The existence of ChatGPT does not predict a future Artificial General Intelligence. That's not to say that some AGI could possibly be released tomorrow. We just don't know. I agree that we shouldn't downplay the potential (both positive and negative) effects on society from a potential AGI or even many kinds of narrow deep learning models, but at this point we're not looking at thinking machines.

 

[Soothsayer786]

As someone who writes for a living, I find ChatGPT to be both fascinating and utterly terrifying. This thing is going to take my job in a few years. It's incredible what it can come up with. I'm thinking about starting a new career as a writer of children's novels. I feed the system inputs, it comes up with the good stuff.

Here is "my" first story! I tried to make it as random and challenging for the AI as possible. It was created using the following input: "create a children's story with a duck a cow a spider a turtle a penguin and an angry caveman"

The title is: "Friendly Caveman Adventure Story"

You can read the whole glorious adventure here:

Spoiler

Once upon a time, in a faraway land, there lived a duck named Daisy. Daisy loved to swim in the pond and play with her friends, a cow named Clarabelle, a spider named Sammy, a turtle named Tommy, and a penguin named Polly.

One day, while they were all playing together, they heard a loud noise coming from a nearby cave. Curious, they went to investigate. Inside the cave, they found an angry caveman named Carl.

"Who are you and why are you making all that noise?" Daisy asked Carl.

"I'm Carl, and I'm angry because someone stole my favorite rock," Carl replied.

The friends were confused. "Why would anyone steal your rock?" asked Clarabelle.

"I don't know, but I'm going to find out and make them give it back," said Carl, grumbling.

The friends felt bad for Carl and offered to help him find his rock. They searched the cave and found nothing, but as they were about to leave, Sammy the spider remembered that he had seen a group of mischievous monkeys playing with a rock that looked like Carl's earlier that day.

The friends thanked Sammy and went to the monkey's tree. They saw the monkeys were having fun and playing with the rock. The monkey's explained that they had found it on the ground and thought it was a toy. They gladly returned the rock to Carl, and as a thank you, Carl invited everyone to a caveman-style feast.

The friends had a great time at the feast and from that day on, Carl was not an angry caveman anymore. Instead, he was grateful and became friends with Daisy, Clarabelle, Sammy, Tommy, and Polly. They all played together and went on many more adventures.

The end.

Carl the Caveman really grew as a person through this adventure. And who could forget those mischievous monkeys? I'm going to be rich. RICH!

 

[Voxera]

ChatGPT is effectively a peek at what is coming. Given something like this can quickly make google think their dominance is challenged you can bet the farm that we will see several versions of this from around the world, and some with features that even the creators cant fully explain why theirs includes it, or why they cant recreate it with consistency.

Remember these systems are black boxes we feed data, and determine they work only based on output testing. We cant understand let alone see these weights used. Perhaps future AI models will allow this, but these and ones before it have no internal transparency.

I think ChatGPT still have one obstacle.

If I remember it correctly it can only answer things up to about 1-2 years ago since that is the data it is trained on, it does not have a normal search engine backend nor does it use other search engines.

That means its useless for searching for current data like news or latest changes so far. Correct me if I am wrong on this.

So while the technology is fascinating it it not really a competitor to google at the moment.

But if they can manage to either make the model up-gradable with more data on the fly, or use it to understand and reformat a query to use in a normal search engine DB, or even use multiple sources of data, then this will be a game changer in how we search the net.

Being able to refine the search as we go along, use the results to fine tune it, that would bring this quite close to the use pattern for the "computer" in Star Trek or Jarvis in the early Iron man (not all the way, but close enough to be interesting)

 

[kinpin]

I

Google's gotten too big and unfortunately... evil.

A punch in it's 149 billion per year empire is good and about time.

I wonder how such a case will play out in court.

If I write a book on Topic X , does that mean every author whose book I’ve read can sue me for damages ? After all most of what we know is knowledge built on from what we’ve read prior.

 

[cbreak]

DeepMind has been making some actual world-changing stuff with their approach of solving games and then (last year) mostly-solving protein folding. It's not like Google is completely behind, they're just not doing the flashy-but-problematic fake-text and fake-image generators.

I'm with the poster that compared this to the self-driving-car craze of the mid-2010s or the voice assistants of the 2020s. Everyone thought they needed to get in on the "next phase of transportation" that was going to arrive in 3 years, and then it turned out replacing a $10/hour taxi driver with $100k of sensors didn't make sense. As long as image generators are making people with 7 fingers per hand and text generators are extolling the virtues of eating broken glass, I'm not seeing where the money is coming from.

Google is also doing the flashy stuff. After all, with Waymo, they're the undisputed leader in self-driving cars, and they have Lambda, Imagen and the like.

https://imagen.research.google/

 

[AbboFun]

…
And then one day you find ten years have got behind you
No one told you when to run, you missed the starting gun
…

 

[Oliver Weichhold]

"20 new products" .. which get cancelled a year later.

 

[Tall Dwarf]

As someone who writes for a living, I find ChatGPT to be both fascinating and utterly terrifying. This thing is going to take my job in a few years. It's incredible what it can come up with. I'm thinking about starting a new career as a writer of children's novels. I feed the system inputs, it comes up with the good stuff.

It sounds like you are planning to switch from a job that AI will replace in a few years to a job it replaced a few weeks ago.

How much money in it really? Kids have nothing to read? All Dr.Seuss books turned out to be offensive?

 

[uhuznaa]

I could understand that Google must be extremely enticed to have an AI trained on all the Internet data they have readily indexed nearly in real-time in their databases. This would lead to something you don't really "search the Internet" with but something you just ask a question and it delivers a single answer. And THIS is terrifying if you think about it but I guess most unsuspecting people would just love it... which makes it even more terrifying.

Wait for it, in a couple years this will be the standard free Google interface (integrated with ads and paid-for recommendations if looking for products) and a "real" honest search interface will only be available with a Google Pro account for $50/month. It's basically inevitable.

 

[macosandlinux]

I asked ChatGPT to make a story on how Aristotle was a marmot. Here's the result:

Aristotle Marmot Secret Story

Once upon a time, Aristotle was a brilliant philosopher and scientist. But he had a secret that he kept hidden from the world: he was actually a marmot in disguise.

Aristotle had always been fascinated by the natural world, and as a marmot, he had a unique perspective on it. He spent much of his time studying the behavior and habits of the animals around him, and he used this knowledge to inform his philosophical and scientific writings.

However, Aristotle knew that if the truth about his true identity were to be revealed, it would cause a scandal and ruin his reputation. So, he lived his life in secret, pretending to be a human and using his knowledge to make groundbreaking discoveries in the field of science and philosophy.

Despite the challenges he faced, Aristotle remained determined to continue his work and to make a lasting impact on the world. And in the end, his contributions to the field of science and philosophy were so significant that they far outweighed any doubts or questions about his true identity.

In fact, even today, many experts believe that Aristotle's unique perspective as a marmot was one of the key factors that contributed to his success as a philosopher and scientist."

OK, so what if we change that to lizard people and pizza parlor /s

 

[Panashi]

"its better late than never"

And i really liked how concerned they are of the cons of bringing up A.I., Cause no matter which ever company brings the better one, No one can earn the trust of users like Google does!

 

[Numfuddle]

It also depends on how your business operates and how your market is structured how disruptive AI bots like ChatGPT will be and how soon they'll be able to disrupt your business.

An example: A lot of my friends are lawyers (one of my larger failings as a human) and because of the recent hype - and also because law firms dreams involve getting rid of most of the associate lawyers and legal clerks - they all tried those AI bots.

They all claimed that the AI was competent enough to come up with a customized version of a draft contract or draft agreement even for a "complex" topic like a merger or company purchase - so essentially it created all of the custom boilerplate - that could then be refined.

The legal community places a lot of importance on this because - similar to big SW companies like Facebook or Google - they want to rid themselves of as many of their employees as possible and replace as many of them with automated systems. I had more than one senior associate tell me that new hires now have the impertinence to ask about things like "work-life-balance" and "benefits" and how the weaklings no longer want to work 14 hours a day, 7 days a week and are no longer content with being treated like serfs just for the promise that they might become a senior associate one day and make a lot of money some time in the future (a lot of big law firms are essentially pyramid schemes just with more back stabbing)

But I digress.

What they want is to replace all of the low value add work and the "boilerplate" part of drafting up legal documents with automated "AI" systems in order to remove most of the entry and middle levels of staff and they expect to still be able to sell the "individual" parts of their service to customers (they expect that people will still want to hire seniors from competent local law firms for the experience and expert knowledge)

In businesses like that (similar to insurance or banking) AI bots might disrupt sooner and more significantly similar to how writing departments, secretary staff, technical writers were replaced by word processors or how technical drawing departments have been replaced by CAD/CAM SW packages and are now done by engineering as part of their workflow

 

[kinpin]

If Google was worried about copyright then Google News wouldn't exist.

Like all big incumbents, they have a great revenue model that makes them tonnes of money. Reinventing the wheel on that is not what big companies do well, whether Big Tech or otherwise.

A much better analogy will be 3D TVs because self driving cars is actually around the corner.

 

[leukhe]

Rememeber this story?
https://meincmagazine.com/tech-policy...o-claimed-lamda-chatbot-is-a-sentient-person/
openai solved this by letting the AI response telling it "As a AI i don't have an opinion"instead of the google internal response "you want me to tell i am a person, i tell you i am a person"

 

[entropy_wins]

On a side note a hack project adding Clippy (Clippit) as a UI for some of the new AI would be amazing on all levels

Miss Minutes?

 

[germanofthebored]

Google's gotten too big and unfortunately... evil.

A punch in it's 149 billion per year empire is good and about time.

Honestly, with a technology like AI I would be quite happy if the pace is set by some company that is big and conservative enough to take it slow. A small and nimble company that follows the "Move fast and break things" mantra makes me a bit nervous

 

[akenthet]

I can definitely see why Google is worried.

I tested ChatGPT at work, by asking it to write a technical report for me. It looked very good, for someone not very familiar with the field. But it was very empty of actual, useful information.

Still, it was impressive, fun to test, and I might use it to create general background info text in the future.


But I don't feel that my employment is under any theeat... Yet.

This is similar to my test with it on some technical writing. Superficially, the text looks alright. Maybe at the level of a high school student. I was impressed and am curious how fast the future progress will be.