Here's the weird thing though, as discussed yesterday [1], the Microsoft demo was totally broken as well. In the end, everyone is looking for a big platform shift (next MOBILE!!!) and hype getting ahead of reality. When we look back in 3 months we will wonder why we thought LLM/GPT type queries will replace search in all forms.<p><a href="https://dkb.blog/p/bing-ai-cant-be-trusted" rel="nofollow">https://dkb.blog/p/bing-ai-cant-be-trusted</a>
Xoogler here.<p>I'll ignore the AI and "has Google lost its way?" threads, not that they're not interesting.<p>Rather, Sundar: he's the inevitable product when you hire a CEO based on his longevity and whether everyone likes him. In the military they distinguish between a "barracks general" and a "combat general." He's the former. He looks good when nothing bad is happening.
When Sundar doesn't react fast enough: <i>Google's gone. They no longer have the ship-fast mentality of an upstart. Perfect is the enemy of the good. The ethics lords have taken over Google Brain. Google needs a war-time CEO.</i><p>When Sundar reacts fast enough: <i>Google's gone. The product was rushed. It is no where close to perfect. Could it be Google's been bluffing about its AI. What a knee-jerk reaction. Google needs a war-time CEO.</i><p>Meanwhile, Satya in Redmond: <i>I want people to know that we made [Google] dance.</i>
(Disclaimer: Google employee, I don’t work on Bard, views my own, etc.)<p>I don’t think either Bing Chat or Bard is ready to be released widely. You need a certain mindset to be able to wring value out of them, and most normal users will not think this way.<p>That being said, it’s probably fine to release these tools to enthusiastic early adopters who understand how to use them.
Whats with everything being a dumpster fire nowadays? how about a simple hot mess? or what happened to a good ol hell in a hand-basket? or Ninth Circle Of Hell? I'm sure there are better euphemisms than the dumpster being on fhir as that indicates someone stoking it whereas the reality is more indecision and inaction causing confusion and demoralizing the teams.
> Google employees<p>> Interactive chart of Alphabet (GOOG) annual worldwide employee count from 2010 to 2022. Alphabet total number of employees in 2022 was 190,234<p>There should be a rule where you can't over-represent a group of people in media headlines.<p>What percentage of 190,000 people called the CEOs response a dumpster fire?<p>What percentage of 190,000 are even aware of ChatGPT or Google's response to it?
People hating on Sundar completely overlooking the fact that he pivoted Google to an AI first company 6 years ago, way before AI was hyped up. <a href="https://venturebeat.com/ai/ai-weekly-google-shifts-from-mobile-first-to-ai-first-world/" rel="nofollow">https://venturebeat.com/ai/ai-weekly-google-shifts-from-mobi...</a><p>He funded the Brain team that came up with major research breakthroughs like Transformers that power ChatGPT.<p>AI is embedded behind the scenes in pretty much every Google product, you just don't know it because it's not packaged into a fancy chat interface. They didn't push Bard because they know LLMs have problems with getting facts straight, and they still are rolling it out in response to ChatGPT very cautiously.
From outside, I found it to be a dubious move myself.<p>ChatGPT went first, nothing Google does will change that, that's too late. The only thing they can do if they don't want to lose face is the "we didn't make it available because we want to do it well" strategy. It has been used many times by Apple when competitors beat them at some features, so much that it has become a meme at some point, but it worked.<p>And I've seen hints of Google going in that direction, pointing out things like reliability issues most people came to know about by now, implying that they already can do what ChatGPT can do, but unlike their competitors, they don't release half-assed products.<p>But their latest announcement breaks everything, they are essentially admitting defeat. They didn't manage to be the first to market, and they can't claim they took their time to release a quality product instead. So it was either a terrible communication blunder, or they really didn't have much choice, with the latter being a real possibility, I understand investors reactions.
Seeing a lot of criticism towards Sundar lately, and it's not without merit. There is blood in the water now and I wouldn't be surprised if his time is limited.<p>He is good as a caretaker who didn't rock the boat, but Google needs bold innovation and competence if it intends on staying relevant in the years to come.
I am super excited about ChatGPT even just yesterday it helped me fix a really difficult bug. (I was so excited that I even wanted to pay for it but their payment was broken haha)<p>However, I am starting to think that moving it to bing already wasn’t a smart move. These things are great on OpenAI scale. But I fear they are a money pit with bad pr for google and even bing scale.
I’m struggling to find the link because of all the new coverage, but Bard was mentioned as a Google project in <i>2019.</i> It had a prefix in the name though, which I'm struggling to remember (it wasn't OpenBard, but it was SomethingBard).<p>If that is real (and hopefully I can find it and update this post)… that would indicate Google has been working on it for years, never thought it was ready, and then decided to snap-deploy it after ChatGPT’s success. Which, predictably, would make a massive mess internally as reported here. I find it believable.<p>UPDATE: OK, it was called "Apprentice Bard." However, that names comes from CNBC in 2023, but CNBC doesn't say when "Apprentice Bard" actually began development. However, "Apprentice Bard" was a replacement for Google's first internal chatbot solution <i>Meena</i>, which Google did deploy internally and also publicly revealed only for bragging rights (and not actual use) in January 2020 (and so the 2019 figure in my head does somewhat line up). So if you think about the Bard project as having started with Meena... sounds about right. <a href="https://ai.googleblog.com/2020/01/towards-conversational-agent-that-can.html" rel="nofollow">https://ai.googleblog.com/2020/01/towards-conversational-age...</a>
"And above all things, a prince ought to live amongst his people in such a way that no unexpected circumstances, whether of good or evil, shall make him change; because if the necessity for this comes in troubled times, you are too late for harsh measures; and mild ones will not help you, for they will be considered as forced from you, and no one will be under any obligation to you for them."<p>Sundar is the Ballmer of Google, and maybe the Jassy of Amazon.
I wonder if this is going to end up like Waymo vs tesla. Remember when people were criticizing waymo for going too slowly (tesla expects to put level 9000 self driving in millions of cars by the end of <five years ago>!!)? But it turned out that doing it in a way that the market requires was slower than commentators expected, and waymo seems to still have a great shot at being the leader.<p>Or maybe it ends up like uber vs lyft. The cost of using LLMs is so high, maybe we see products with unsustainable economics in the quest for dominance, followed by something shittier.<p>Personally, I'm in the camp that google is in a great position specifically because of their TPUs. They already have the infra to do this at crazy scale <i>including the hardware</i>. I'm betting it's a marathon, not a sprint, before we see an economically sustainable LLM based interact-with-all-of-human-knowledge tool like we're expecting. ChatGPT has spoiled our expectations by being free.
Tech reporting is so hysterical these days, and so is social media of course. A crappy demo, and before you know it, Google is declared dead. AI has wrong answers, which is the end of the world. Plus it's racist because I tried for 3 days to trick it into saying something racist.<p>Drama is the product, not reality or any actual in-depth tech journalism.<p>Google got a wake-up call, but search will be just fine. Google controls search defaults on Android, iOS, Chrome on Windows, and Firefox. Not to mention on TVs, voice control, it's everywhere.<p>Google can launch a new product to billions of people with the press of a button, Microsoft...not so much. You could argue they can push it into Windows, but most people are on their phones these days. Microsoft has no serious presence in the consumer space anymore.<p>This is what will buy Google time to integrate a competing service, which I expect to be better as it sits on a lot more data.<p>Just kidding, I would absolutely recommend dumping Google stock.
I hope this helps shake up Google's product marketing which hasn't met expectations. Poor segmentation, poor GTM, poor messaging, poor community engagement , poor CTAs on marketing messages.<p>Everyone's aware of the messaging-app missteps. I'll share a more tactical example. Recently Google Home "beta" was launched with lots of messaging via email & twitter to join the Beta. All of the CTAs were broken. The "beta" that was expected in days took a couple months to land. Compare that to Bing ChatGPT Beta which had a clear CTA into the enrollment program ( activate Edge, Bing, etc) and waitlist.<p>Google has a strong community of fans but they do a poor job of engaging their audience. Microsoft by comparison has MSDN, MS Insiders, etc.<p>The Bard (horrible name, btw) debacle was not a singular event but a culmination of many product marketing issues. As a fan, i hope they make some big changes.
If I were IBM management, I'd be getting a laugh out of all of this. When IBM had to scale back their ambitions for Watson, especially those related to automating medical diagnoses, they were roundly mocked. Do not ask, dear tech companies, for whom the bell of AI overpromising tolls - it tolls for thee!
There will always be people who think they knew better. It's not the end of Google or their Bard. It's an emotional outburst precipitated by lousy HR numbers. People are fearful and need to lash out a bit. Ars is riding the slight wave of disenchantment among a few emps.
Whatever happened to move fast and break things? Who cares if it wasn't perfect in the first iteration. It was a shot across ChatGPTs bow, and pleased the shareholders.
Does this make Alphabet stock a buy? In general I have a rule to never invest in adtech-based companies but a 12% discount is attractive.<p>Edit: a quick look at the chart shows the drop was off a recent peak and current price is roughly the 3mth mean. Tempest in teacup then.
The writing is on the wall. Sundar hasn’t done anything but maintain the status quo at Google. That is a loss for google, particularly GCP didn’t gain any market share, couldn’t productize their self driving car platform effectively, the ad market is facing increased pressure from all governments. I see him replaced soon or some activist investor demanding some changes.
Google has lost its way since they have switched their own mantra from "user-first" to "AI-first".<p>On the flip side, I think Microsoft has made a strategic mistake of going after search openly. Satya literally called out Google to compete.<p>Google also has a lot of money and will not lose easily. This war is going to cost a lot of money for both of them.
We changed the URL from <a href="https://arstechnica.com/gadgets/2023/02/google-employees-criticize-ceo-for-dumpster-fire-response-to-chatgpt/" rel="nofollow">https://arstechnica.com/gadgets/2023/02/google-employees-cri...</a> to the article that one is pointing to.<p>Submitters: "<i>Please submit the original source. If a post reports on something found on another site, submit the latter.</i>"- <a href="https://news.ycombinator.com/newsguidelines.html" rel="nofollow">https://news.ycombinator.com/newsguidelines.html</a><p>Recent and related:<p><i>Google employees criticize CEO for ‘rushed, botched’ announcement of Bard</i> - <a href="https://news.ycombinator.com/item?id=34758402" rel="nofollow">https://news.ycombinator.com/item?id=34758402</a> - Feb 2023 (36 comments)
With 150,000 employees, you can write an article saying "Google employees say X ..." regarding pretty much any subject under the sun. So why was this particular article written, which refers to .002% of the company?
I could not believe Sundar and the team around him for a simple unforced error:<p>"That’s why we re-oriented the company around AI six years ago"<p>I suppose this was meant to reframe the conversation around Google being a leader, but I think it has the opposite effect. It comes off as defensive and of course, quite honestly, if you planted the flag so much earlier, why don't you have a product to show for it in the way that OpenAI does?<p>What's perhaps more interesting though: does Google have an innovator's dilemma here?<p>Google built a flywheel that drives its ads business: sites are hungry for traffic so they structure in such a way that attracts Google's attention, and that structure in turn helps Google serve relevant ads. People tend to click on the first link (the most relevant link).<p>In the context of a chat interface, what happens if there is no link? Or only one result (the perfect answer)? What if there is much less traffic going directly to websites? And yet without that long tail of websites, how do you train your language model to know the answer?<p>Fascinating times!
Google Prepared Transformer => GPT, that has a nice marketing ring and clearly claims the invention/innovation. Google took their eye off the ball and fumbled. A "goto market" division could have helped the inventors/innovators.
The blowback about googles response is unreasonable. Everyone is making some wild assumptions about gpt-based search. We don’t even know yet if it’s fit to replace a search engine.<p>Why not wait for a moment and see what unfolds? ChatGPT has only been alive for a few months. Why does google need an answer immediately? They should take their time and be calculated instead of rushing out some half baked AI search solution.<p>How is GPT-search supposed to be profitable anyways? The traditional model of targeted advertising would be insufferable if injected into a chat bot.
I think LLMs are great, but most companies have not figured out what they are truly useful for. Hint: it's not for search. LLMs are not knowledge databases, and should not be relied on for unknown information. In my opinion, they are more useful 24/7 editing service. Write a paragraph and ask it to paraphrase. Ask it to make a new story that can be edited. If you don't have any idea what the answer to the thing you're asking is, you shouldn't rely on the response from the bot.
Will anyone remember how Bard was released in 5 years, if it delivers on its promises? I doubt it, so the hand wringing over a stumble this early seems overblown.
I think people are too soon to make judgements. Zuck going all in on the metaverse might be one of the most brilliant and daring business moves of recent history, or it might be a poorly implemented dream of chasing Apple.<p>Google being 'second mover' to AI Chat might be the smart play, or it might be terrible.<p>I'm very surprised even here people aren't realizing we don't know how this plays out yet.
Left out of this is that Google is trying to promose a perfect product, and trialing it out only internally<p>Microsoft shipped someone else's product to the world and is getting the results to iterate on for their next version.<p>The fumbled response by Google distracts from the fact that you're going to need exponential amounts of training data and only one of the companies is in a position to collect it.
ChatGPT is neither an earthquake nor a tectonic plate faultline. Along with ever improving algos in nearby domains (images, speech etc) it is a signal, of sorts, of the slowly accumulating energy that will, eventually, unleash the next phase of information technology.<p>It is not an accident, imho, that it is Microsoft that somehow emerges as Google's "nemesis". After all it was MS' business model that Google disrupted with mass market "freebies" such as gmail, docs etc. While "disrupting search" might be the immediate skirmish, the bigger battle is about broadly defined "AI augmented information management": who will offer it to the masses, under which business model etc.<p>My feeling (nowhere near a complete analysis) is that Google will lose this war even if, as it is widely assumed, it has the technical advantage. This is because it has been happily cornered in an extremely lucrative but ultimately dead-end business model. AI augmentation as part of its existing search/adtech model is a marginal and even problematic (loss making) addition, whereas as part a re-invented MS it can blow new life into its old bread-and-butter business that has been commoditized. A wildcard in all this is (as always) Apple, but whatever its eventual approach to AI, it is likely to be just another headwind for Google.<p>NB: I am a lifelong open source enthusiast, I have no stake or love for any of the above entities. The ideal, least dangerous, most beneficial form of AI augmentation would be as a widely available and open source digital public good.
They're the guys who built it. It's not like he's demoing something that they spent 4 days building. It's been in dev for 4 years. Everyone knows Google is primarily a BD play right now. The engineering there is in the "we'll make the button blue by next quarter" quality.<p>But BD plays build quite good moat. So I think they'll be fine.
Gpt as everyone that uses it extensively by now only excel in certain tasks. There is a lot of bias towards it being “great” as many of its answers look really plausible and not verified because what’s the point of using Chatgpt if I have to google again to verify?<p>In my view this is overhyped as a google replacement and the cracks are showing.
Instead of focusing on adding LLM to search, they should have rolled it out as a GCP service. There would be less reputational risk and it would give them a chance to learn how it would be used when it did eventually integrate into Search.
I am constantly amazed that chat bots so completely capture imagination.<p>They have been, and will always be a hollow shell of interaction. Has social drive atrophied so much that many (most under 30?) prefer to talk to nothing instead of taking to someone?
Someone somewhere will call this the lost decade of Google. But they will bounce back i think. The current reaction is impulsive, not surprising for the internet age.
Remember when winner of the voice interface war was going to win the future of computing? Apple has a "dumpster fire" of a voice assistant and they're still around.
as a developer I ran few tests on ChatGPT for example "write me a javascript code that does canvas animation" it was amazing to see it write actual working code. but it took about a minute to write it. If I had "Googled" the same thing, in under a second I would have found multiple working code examples (thanks to sites like StackOverflow). IMO, ChatGPT has a long way to go.