Google CEO Issues Rallying Cry in Internal Memo: All Hands on Deck To Test ChatGPT Competitor Bard (cnbc.com) 59
Google CEO Sundar Pichai told employees Monday the company is going to need all hands on deck to test Bard, its new ChatGPT rival. From a report: He also said Google will soon be enlisting help from partners to test an application programming interface, or API, that would let others access the same underlying technology. The internal memo came shortly after Pichai publicly announced Google's new conversation technology, powered by artificial intelligence, which it will begin rolling out in the coming weeks. Google has faced pressure from investors and employees to compete with ChatGPT, a chatbot from Microsoft-backed OpenAI, which took the public by storm when it launched late last year.
"Next week, we'll be enlisting every Googler to help shape Bard and contribute through a special company-wide dogfood," Pichai wrote in the email to employees that was viewed by CNBC. "We're looking forward to getting all of your feedback -- in the spirit of an internal hackathon -- more details coming soon," he wrote. Microsoft is reportedly planning to launch a version of its own search engine, Bing, that will use ChatGPT to answer users' search queries. Microsoft is holding its own event Tuesday with participation from OpenAI CEO Sam Altman. "It's early days, we need to ship and iterate and we have a lot of hard and exciting work ahead to build these technologies into our products and continue bringing the best of Google Al to improve people's lives," Pichai wrote in his note to employees Monday. "We've been approaching this effort with an intensity and focus that reminds me of early Google -- so thanks to everyone who has contributed."
"Next week, we'll be enlisting every Googler to help shape Bard and contribute through a special company-wide dogfood," Pichai wrote in the email to employees that was viewed by CNBC. "We're looking forward to getting all of your feedback -- in the spirit of an internal hackathon -- more details coming soon," he wrote. Microsoft is reportedly planning to launch a version of its own search engine, Bing, that will use ChatGPT to answer users' search queries. Microsoft is holding its own event Tuesday with participation from OpenAI CEO Sam Altman. "It's early days, we need to ship and iterate and we have a lot of hard and exciting work ahead to build these technologies into our products and continue bringing the best of Google Al to improve people's lives," Pichai wrote in his note to employees Monday. "We've been approaching this effort with an intensity and focus that reminds me of early Google -- so thanks to everyone who has contributed."
Re: (Score:2)
Dear Google (Score:3)
Fix your f*cking search engine first. ChatGPT is flash in the pan.
I'm about ready to roll my own search engine.
Re: (Score:2)
Please make it one with blackjack and hookers, I'm tired of that censored shit.
Re: (Score:1)
Looks like you'll have to switch to Bing. [slashdot.org]
Re: (Score:1)
> Fix your f*cking search engine first.
They are a near-monopoly, they don't have to. They can suck without consequences.
Re: (Score:2)
Mod parent up, though I think the term "monopoly" has mostly been redefined to have no meaning. Bork cleverly redefined the term and thus limited the scope of monopoly to certain narrow economic abuses. And (of course) the abusers gladly (slightly indirectly) funded the judges who bought the BS.
Monopoly should be defined in terms of loss of choice and suppression of competition and innovation. The only thing that surprises me about this story is that the google wasn't clever enough to buy ChatGPT before it
Re: (Score:2)
I think MS might have called dibs...
Re: (Score:2)
I don't think MS really deserves much credit for that part of it. Niche monopolization or dominance has a long and dishonorable history. ATT or GE are probably the leaders in terms of stifling innovation. But you might have another candidate?
I only credit Microsoft with one innovation, the EULA approach to eliminating liability, which has become standard in a bunch of niches and all over the Internet. I used to give Gates credit for the idea of marketing above the end users, but now I think that was just a
Re: (Score:1)
is that the google wasn't clever enough to buy ChatGPT before it became a problem.
Google apparently either had its own in-house ChatGPT-like project, and would have had many bid competitors to anything nascent. Microsoft probably won that round before ChatGPT was released.
What's incompetent was that the CTO wasn't keeping on top of the competition. Maybe its me, but I think its pretty stupid strategic thinking to try to push out something that may only be 95% complete in a cutting edge field like applied AI. Tech is tech; either you already had something ready to release, or you're ju
Re: (Score:2)
I'm confused - can you explain what's wrong with Google, relative to others?
Combination of things (Score:2)
I'm confused - can you explain what's wrong with Google, relative to others?
A combination of things.
Firstly, SEO [wikipedia.org] is a thing, a big thing, and it has driven every large website to pad their site with innumerable pages of shallow information. Things like "these 5 exercises help reduce knee pain", "the best 5 video editors for linux", and so on.
Secondly, everything is on social media apps these days, these sites want to keep you on their site, so the information isn't text in the traditional sense - clicking on a link goes to a javascript function that constructs a new page of informa
Re: (Score:3, Interesting)
They can't. They're backed in to a corner by ad revenue. The reason you see so much SEO'd crap when you search is that it causes more ads to be displayed, and the reason you can't even find stuff that doesn't have ads (like my old archived Slashdot posts, which used to show up and don't now) is because they don't generate revenue. They've turned the coffin corner. Google search is a giant that just took a poison dart and is still walking. To most people it looks fine, but to anybody who's really paying
Re: (Score:2)
ChatGPT may or may not replace Google, but something will and that something will use a model that's less ad-driven, or harder to game.
ChatGPT is not going to replace Google Search, as is. The only reason Google Search is in its pre-eminent position is that Google most cost-effectively invests in web crawling, and then finding ways to process their crawler data warehouse, while structuring ad services to help narrow desirable web results, and then finding the most cost effective way to set charges for advertisers. I don't believe there's a single thing about Google Search that cannot be replicated by a competitor; they just don't have th
Re: (Score:2)
But if people are just asking AI for answers, where's the incentive for people to churn out the web pages that AI learns from? Pages that no longer get clicks?
I don't love that the web turned out to be ad-driven, but it's because people chose that over any other option such as subscriptions or micropayments.
Re: (Score:2)
Dude, just roll your own out so we can use it!
They would but (Score:2)
Good luck (Score:3)
Re: (Score:2)
Re: (Score:2)
The human-input reinforcement learning that separates ChatGPT from regular GPT is exactly what you get by having a whole bunch of people (you trust to not teach it to be a Nazi) talk to the thing.
Comment removed (Score:4, Insightful)
Re: What about "No"? (Score:3)
Re: (Score:2)
Isn't the whole concept behind Bard that, unlike ChatGPT, it has realtime access to data?
Re: (Score:2)
Time for the AI apocalypse? Superintelligence was the book? Pretty sure Bostrom was the author.
Re: (Score:2)
Re: (Score:2)
It'll make it at least as accurate as a Google search, and in practice probably far moreso, so what exactly is your point?
Re: (Score:2)
Re: (Score:2)
A Google Search "simply repeats things from the Internet without understanding it or giving context". And indeed, an AI system is FAR more likely to understand the context than a simple text search.
Re: (Score:2)
Fucking scams (Score:5, Interesting)
You are doing nothing but training a billionaire product for free so they can later disable the free access and charge you an arm and a leg for. And prolly a kidney.
These are terrible as long as there a profit motive behind them. These need new data to grow.
Corp will force this on everyone. You will only be able to interact with your computer indirectly thru the bot.
The bot won't be able to create nothing new. It can only use its training data.
People are willing to completing trust this thing.
So eventually, the only input into the system will be the training the computer already did. Nothing new will be made. And we rot.
Re: (Score:2)
What you give them: input
What you get: output
Is the output worth the input? That's subjective. But you're certainly not getting nothing.
Before ML you were training corporations, now you're training software for corporations. The difference is negligible to the user.
Re: (Score:2)
But no one will know how to fix it. All they will know how to do is ask the computer. And the computer is just as clueless.
It smacks of desperation (Score:1)
Emergency! All hands on deck! Drop everything, this is no drill! "we'll be enlisting every Googler".
What previously back-burner project do we have in the basement that we can shove out there and call it competitive?
Google is very much afraid that this generative AI stuff will get away from them and their wide moat will be breached.
Well almost all hands (Score:5, Funny)
Seems like they are short about 12k hands they could be using to test Bard...
Unless they make laid off employees work with Bard to answer questions about benefits and health care I guess, maybe that's why they were let go.
Pichai (Score:1)
Re: (Score:2)
Google is already running it, just not in an easy-to-use fashion. Any tech nerd can actually run their own GPT-like model on Google Cloud in a few minutes/hours. There are plenty of them out there that can generate NSFW stories (because off course, the Internet is for porn).
The only thing that changed is that OpenAI put a GUI and a ton of marketing on it.
Re: Pichai (Score:1)
And a backend that everyone else will be copying?
Are natural language AIs that useful? (Score:3)
Re: (Score:1)
Re: (Score:2)
Re: (Score:2)
I'm sure this tool will be used as much as search engines by everybody.
No, this is replacing the search engine and all search functions that a corp can stuff it into.
Then charge you to access it. What are you going to do about it?
Re: (Score:2)
If you look at how a natural language AI works, it's basically taking a sentence up to a certain point and guessing (using statistical probability) the next word in the sentence based on everything in its database. It's useful to speed up auto complete, but if you try to use it to write something interesting it'll just spit out something that looks about right, but without regard to the meaning. It's just taking advantage of the fact that a good portion of our communications are fluff and niceties, like "sure is hot today, ain't it?" and "talk to you soon!" Even if you train it on stuff Newton and Einstein said, you're not going to get any new theorems out of it because it's not actually thinking, and doesn't "know" what it's talking about. Sure, we can replace the person who talks to you at the drive thru, but we had apps to order our food now anyway. Using it to do your homework is just shooting yourself in the foot. What problem is this going to solve?
Coding: The code generation AI's are fairly useful. There's a lot of generic blobs that it can spit out and let coders do the more interesting things.
Customer Support: I've spent literal hours trying to accomplish something fairly simple with Samsung's tech support, spending tons of time on hold, and much more time dealing with humans who weren't helpful because they couldn't grasp the vast complexity of Samsung's customer support system. Maybe a good AI could handle 90% of interactions and let better-train
Re: Are natural language AIs that useful? (Score:1)
Living cells have very little to go on except their relation to the next closest cell. So statistically they are just playing the odds something more than just random pattern matching. So maybe the scientists and engineers are on to something? And, yes replacing human to human contact is what we need to be wary of. Technology is supposed to give us more time to do so not compete with us for it.
Re: (Score:3)
If you look at how natural language works, we all process on "next word".
Hofstader put this best in the intro to Surfaces and Essences:
And so, in summary, a conversation constrains the ideas in it, the ideas constrain the sentences, the sentences constrain the phrases, the phrases constrain the words, and finally, the words constrain their letters.
It reminds me of the Python skit involving "No Time To Lose", and no one understanding what that phrase means. "Toulose as in France?"
It's about context. The "m
Re: (Score:2)
We can apply procedural logic.
You could train a transformer on all the induction puzzles and solutions in existence, but pose a new one in natural language and it's still going to be clueless.
User guided training is all for these systems (Score:3)
All these early embarrassments, many of them public as I said (but probably a lot of them are not), had helped refine the GTP3 and ChatGPT products into a more usable state.
Also, they know what kind of things the users want the most, what things fail the most and precisely what to do for improving their next product.
That is a significant body of internal knowledge Google has yet to develop. Despite their historical fretting about IA's toxicity and ethic compliance, they really haven't had the chance to test it out where it matters.
Unusable state (Score:2)
OpenAI's product had the advantage of having submitted their product to a deluge of abuse since a few months back. Some of the mischievous prompts are funny or creepy or just embarrassing, and end up publicized in the social media.
All these early embarrassments, many of them public as I said (but probably a lot of them are not), had helped refine the GTP3 and ChatGPT products into a more usable state.
Also, they know what kind of things the users want the most, what things fail the most and precisely what to do for improving their next product.
That is a significant body of internal knowledge Google has yet to develop. Despite their historical fretting about IA's toxicity and ethic compliance, they really haven't had the chance to test it out where it matters.
This makes ChatGPT less useful, not more.
The programmers of ChatGPT have implemented specific rules to prevent it from presenting certain political positions, apparently using social justice for guidance.
The system will no longer present the strongman argument for fascism, for example, but is happy to present the strongman argument for communism. Check the before [twitter.com] and after [twitter.com] snapshots of the program.
The public needs to see the arguments *for* Fascism/Naziism in order to properly form a political opinion and t
He's jealous, but not of the AI (Score:1)
He's pissed that Chatgpt has collected so many phone numbers. That's *his* specially, dammit!
Behind The Game (Score:4, Interesting)
Google's reaction to ChatGTP makes it clear they were really behind and didn't at all anticipate it. They're scrambling, because they see how it could completely disrupt their business. They issued a 'code red' about it and have an all-hands effort to stop it. This type of approach rarely works well. It's a rush job with the purpose of stopping the potential impact. They're not looking to make the best product for their users, but rather stop users from going elsewhere. The focus is wrong. No one became an industry leader by focusing on being an industry leader, they did it by focusing on making something people actually want.
Google+ All Over Again (Score:2)
Yeah, that ended well...
Big tech still clueless! LOL! (Score:3)
So THIS is what gets Google so worked up for an "All hands on deck!" development push, huh? Microsoft's backed AI chatbot technology did something unique that Google didn't do themselves. So they MUST copy-cat it, at once!
I played with ChatGPT and it's pretty cool. And even better if MS is using a newer, faster iteration of it that's filled with more recent info than the 2021 cut-off.
But it's sad that the brightest minds and biggest players in the tech-space can't just focus on original products and services. I mean, how much effort was wasted at Microsoft to build "Cortana" when it's never been anything but a bad "also ran" copy-cat of what Apple achieved with Siri and what Amazon built out into a much more impressive powerhouse with "Alexa"? (And sadly, Amazon doesn't even seem to care that they made the "best" voice assistant of the bunch, because it's ordering cutbacks in that whole department! Since they can't figure out how to monetize Alexa, they don't care that it works well!)
I just think we'd all be a lot further ahead if each of these companies focused on ORIGINAL product and service R&D -- not even bothering if they're just re-inventing a wheel made by a competitor.
Somebody is late to the game... (Score:2)
... and sounds a bit panicked. Of course, no amount of testing will fix this thing. They will just have some understanding what is broken. But fixing things like this takes quite a bit of time and actual insight and cannot be rushed.
Google assistant (Score:2)
Even if it is something half-baked and rushed, it's got to be better than Google Assistant. The fuckin' thing can't items in my shopping list if they start with a capital letter!
Ignore it (Score:2)
Bad news for research (Score:2)
Corporate science and engineering research, including AI, depends on funding from monopolies. Thus it has always been: Bell Labs, IBM, Xerox, Kodak, Google. Kill the monopoly and you kill the expendable research funding. This slack will not be taken up by MS. Better hope that Google can catch up.