Editor: Editorial Department
Sam Altman's ridicule of Google is about to rush out of the sky! Just now, he said in a post on x that he "can't help but think about the aesthetic differences between OpenAI and Google". At the end of the two press conferences, Google was left behind for more than just a demo. The desktop version of ChatGPT intends to subvert the entire browser search form, and Google's hometown is about to be lifted. Can crazy restructuring save Google from fire and water?
Now, the mutual confrontation between OpenAI and Google is already on the surface.
After two days of the launch event, Sam Altman was clearly pleased with OpenAI's performance over Google.
Just now, he posted on x: "I try not to think too much about our competitors, but I can't help but think about the aesthetic differences between OpenAI and Google." Between the words, there is uncontrollable pride.
The two pictures of the press conference attached below are very miserable -
OpenAI on the left, the interior layout is elegant and stylish, and the real-time demonstration of live people gives people a sense of live impact.
And Google on the right, the stage of the conference is very far away from the crowd, and the visual theme of the whole conference has a sense of outdated interface from ten years ago.
Not to mention the two-hour duration, which directly made the audience fall asleep.
This photo seems to be a metaphor for the old factory that has entered the bottleneck
Indeed, on one side are the researchers wearing sweatshirts emblazoned with the OpenAI logo and calmly and elegantly showing off the effects of the products, and on the other hand, there are performance artists wearing exaggerated four-color robes and jumping across the audience and shouting "GOOGLE".
Netizens cruelly commented: This is the difference between a $100 billion company and a $2 trillion company.
Yes, the four-color rainbow pasted on both sides of the big screen of Google's press conference, as well as the robe on the crazy artist, are all the signature colors of Google.
Having said that, Altman clearly cares deeply about Google's rival.
Some netizens commented on this: The day when every product release is changed to the day in front of Google is OpenAI's greatest respect for its opponents.
And the research engineer of Google Brain sneered at Altman: On the one hand, you don't want competitors, and on the other hand, all releases are synchronized with competitors, you are really "tea".
Netizens restore Altman's true heart online——
Google VS OpenAI
Brother Chochai was interviewed, and every sentence was OpenAI
After his own press conference was preempted by OpenAI, when he was interviewed by the media at the I/O conference, he also had to face the reporter's fire-
Almost all the questions are directly related to OpenAI, one is sharper than the other, and a good interview directly becomes an attack.
"You guys have a good demo, when will you release the product? OpenAI got it done yesterday, right?"
"No, they haven't released it either, it's just a demo. We have a lot of experience in making products, and we know when to get the products right."
"Some people say that Google's strength is in data, but not in quantity, but in quality. What do you think about OpenAI stealing your data?"
"Then you'll have to ask them, our user agreement is clearly written, and we'll take care of it."
How are Google and OpenAI's voice assistants different?
From the live effect of this press conference, it can also be clearly seen that the difference between Google and OpenAI in the voice assistant route.
The AI voice assistant blessed by GPT-4o is more emotional, directly restoring Samantha in "Her", and the humor and shyness displayed on the spot make it more like a human.
When Google introduced Astra, the word it used the most was "assistant", which clearly shows Google's positioning of voice assistants.
Obviously, when it comes to AI assistants, OpenAI is bolder in its attempts, while Google is more cautious.
Google's attitude has also been traced before. It has been deliberately moving away from "Her" type AI for a long time.
Last month, researchers at Google DeepMind published a paper detailing the potential drawbacks of anthropomorphic AI, saying that such assistants could redefine the boundaries between "human" and "other."
Address: https://arxiv.org/pdf/2404.14068
And this, it can lead to some harm. For example, users will hand over important decisions to AI, disclose sensitive information to AI, or rely too emotionally on AI, and so on.
Such results can be catastrophic. After all, what is depicted in "Her" is not a happy ending, but more of a dystopian future.
Obviously, this issue is not within the scope of consideration of the enterprising OpenAI.
At least in terms of presentation, GPT-4o's live demo allowed OpenAI to directly "kill", while Google's Projetct Astra was as always a pre-recorded demo.
On the good and bad of Google's AI conference: OpenAI is putting pressure on Google
In terms of length, OpenAI's 26-minute press conference is probably just one of Google's 12 appetizers.
Google's lengthy and complicated press conference content gives people the feeling that it probably can't catch up with OpenAI in terms of models, so it simply wins by volume and integrates AI into its own package of products.
Indeed, Google has more complex research and planning on how to further integrate AI into search.
In addition to the Veo video tool and Astra's multimodal AI assistant, all of Google's existing products will be infused with AI, including Google Docs, Gmail, and Chrome.
The most eye-catching feature: the Android phone scam phone detection tool, which can warn us in real-time if the person on the other end of the phone is a scammer
Overall, Google is very confident in its presence at the event, and it is no longer the same Google that rushed to catch up with OpenAI at the end of '22 and early '23.
Everyone remembers that within weeks and months of OpenAI's release of ChatGPT, Google was scrambling to sort out its AI business, looking very embarrassed.
Is Big Tech at a disadvantage in the AI space today? In a few weeks, by the time Apple's conference is over, we may have the answer.
ChatGPT desktop kills Google search and enters the post-browser era
The smoke of the two press conferences dissipated, and what left the deepest impression on the audience was probably the competition between the all-round multimodal model GPT-4o and the general-purpose AI agent Astra.
But in fact, OpenAI's real killer weapon to subvert Google this time is the ChatGPT desktop app it released.
It has the potential to challenge Google's dominance on the web and turn the entire Internet upside down.
ChatGPT desktop version, which will soon be available on MacOS and Windows. Through this program, we will be able to quickly access ChatGPT.
In addition, you can also choose to let it "see" your screen to "see what you see", and you can also interact with it in real time like using ChatGPT.
Upload a screenshot directly to ask questions
The desktop app is a clear indication that OpenAI wants to move from the browser to the desktop.
It's clear that OpenAI is trying to unleash a revolution that creates a whole new way of using computers and networks.
This new form is completely beyond the functional imagination of traditional browsers. Browsers, representing the old Internet, the world of websites, the domain of Google.
And this small step from the browser to the desktop, OpenAI, may be a big step forward in the development of human-computer interaction. What OpenAI hopes is to complete a "search revolution".
From now on, we will have an assistant that can do everything for the user, and what the user needs, just say it out loud.
Sam Altman made it clear in his blog post on GPT-4o:
Talking to a computer has never felt natural to me, but that's changed now. As we add personalization, access your information, and assist you in taking action, I really see an exciting future where we'll be able to do more with computers than ever before.
At the same time, in a recent interview, Altman praised GPT-4o for the unexpected joy it brings to his workflow.
When you're fully focused on your work, simply lay your phone flat on your desktop, no need to switch windows or Google searches, and ask GPT-4o questions directly and get instant answers without affecting what you're looking at on your computer.
The ChatGPT desktop app can gain access to the microphone, camera, files, login information, and the user's screen, making it the best place to be a "personal assistant".
With this "assistant", even smartphone users are hard to get attracted to Google and the web.
In this new way of interacting with computers, the World Wide Web has little place. At best, it acts as a provider of information.
In fact, OpenAI has positioned itself as a gatekeeper of the chatbot ecosystem, launching the Publisher Choice Program, which is developing selected media outlets into partners, with the Associated Press, Axel Springer, Financial Times, Le Monde and others already joining the program.
If OpenAI succeeds, it will have a much greater market impact on our information ecosystem than Google does today.
AI领域大V Allie K. Miller 对于GPT-4o加持的ChatGPT桌面应用，感到非常兴奋。
"Oh my God, this is a killer feature! It's basically a colleague who can share your screen from morning to night and never get tired, so to speak. I can imagine people working hours on end."
GPT-4o，本质上已经是AGI？
As for GPT-4o, many people in the industry have also spoken highly of it.
Jim Fan, a senior scientist at Nvidia, said that apparently, OpenAI is eating Character AI's lunch, with almost 100% overlapping shapes and huge distribution channels.
It's a pivot to a more emotional, more individualized AI, something that OpenAI seems to have struggled to suppress in the past.
Developer Benjamin De Kraker also praised GPT-4o, saying that, essentially, GPT-4o is AGI.
If the public sees it as magic, as a virtual "person" who can hear, speak, see, and reason, what else can you call it but AGI?
Developer Siqi Chen was impressed with GPT-4o's ability to render 3D objects based on text prompts.
"In retrospect, this is by far the most underrated OpenAI technology incident," he said.
In conclusion, GPT-4o has been launched for less than a day, and many features have not yet been made available to the public. But judging from the enthusiastic response that has appeared, OpenAI has touched people's nerves.
Google's way of fighting back: restructuring, more restructuring
How can a giant ship with a big tail turn its bow and adjust its strategy to meet the onslaught of new competitors?
In April, Google did one of its favorite things: restructuring.
The timing is no coincidence, after all, Google has lagged behind in the field of AI for a long time.
In response, CEO Pichai has redesigned his leadership team to move faster and facilitate collaboration between previously siloed departments in the business.
To achieve this, he went deep within the company and promoted a group of senior employees and experienced leaders to manage Google's next era.
But even though it's been much more streamlined, Google's management is still much larger than its peers — it has as many as 18 direct reports.
MVP of Google in the field of AI
Today, Demis Hassabis is probably the most important one for Google.
Hassabis has distanced itself from the company for the past 10 years since he sold DeepMind to Google in 2014.
In April last year, in order to combat the threat from Microsoft and OpenAI, Splitwood decided to merge DeepMind and Google Brain into a super team of about 2,600 employees - Google DeepMind.
Hassabis is the head of this new "AI Command Center".
This week, Hassabis' debut at Google's I/O conference is enough to show Google's determination in AI.
Looking back at the past time, Hassabis' power has been constantly expanding.
In last month's reorganization, various AI-related teams, including building machine learning models and building responsible AI, were merged into Google DeepMind.
And this wave of operations has only one purpose – to help Hassabis and the Google DeepMind team move faster towards AGI.
Obviously, the road is not as smooth as it seems.
On the one hand, DeepMind had to abandon projects that were considered too academic or difficult to turn into a product.
Brain and DeepMind, on the other hand, must let go of their long-standing rivalry. After all, not only did the two teams have had friction over how much research should be shared, but they have even gotten to the point of shielding each other's code.
However, under Hassabis's leadership, Google still relied on the powerful Gemini to win back a city in the competition with OpenAI.
The SVP of Equipment & Services has a lot to offer
The merger of the services and hardware divisions shows that Google is increasingly focusing on hardware products such as the Pixel, especially to seize the key opportunity of "AI on Android".
In a memo released in April, Pichai said it would provide Google with a "great opportunity to reinvent the computing platform for the next decade."
Rick Osterloh, who has been promoted to senior vice president of Google devices and services, has nearly 25,000 employees and one of the company's largest divisions — the size of Google's search and advertising division, led by Prabhakar Raghavan, senior vice president, according to internal documents.
However, many employees expressed disappointment with the progress Osterloh has made on hardware.
On the one hand, it has messed up various products, including Fitbit and Stadia, and on the other hand, the Pixel has almost no influence in the mobile phone market.
Fortunately, when Apple tried to add AI features to the iPhone, Siri, which is responsible for the voice assistant, was very across.
This gives Google an opportunity to combine AI and hardware into a more attractive product.
A 20-year-old veteran will take over the search field
In March, Google appointed a company veteran, Liz Reid, to head the generative search business (SGE).
The appointment is clearly a larger personnel change for the company's search team.
While Reid's appointment hasn't received as much attention as recent changes in other companies, her promotion is significant.
Reporting directly to Raghavan, Senior Vice President of Advertising and Search, Reid will lead a major transformation of Google Search.
At this week's I/O event, Google announced that it will be launching a generative AI search experience called "AI Overviews." And introduced that by the end of this year, more than 1 billion people will have access to it.
Reid is a 20-year Google veteran. He joined the company in 2003 and became one of the first members of the Google Local engineering team.
All her while, her work at Google has focused on how to present high-quality and relevant information to users.
In 2021, Reid reportedly took on a more prominent role in Google search and began to focus on developing AI features.
At the time, Google also promoted Cheenu Venkatachary, senior vice president of search quality and ranking, to the position, another signal of how much Google Search is taking AI in the future.
Venkatachary is VP of Product at Google where she is responsible for AI.
Subsequently, this year Pandu Nayak succeeded Venkatachary as the chief scientist of the search team. He was in charge of search rankings and quality for a long time.
As Google's generative AI search capabilities began to roll out, Reid's team was facing a huge challenge:
AI-generated spam has increased, original content has become less visible, and even the company's cash cow, sponsored ads, has been thrown into turmoil.
In addition to its own challenges, Google also faces competition from strong competitors such as Perplexity and OpenAI.
They have developed the ability to integrate search engines into chatbots. And OpenAI's search product, SearchGPT, which has been exposed, will also be launched at some point in the future.
Altman once talked about wanting to create another "Google Search":
"Google search will show you 10 links, 13 ads, and then 10 links, which is a way to find information. But I'm excited that we can replicate a better 'Google search,' and maybe there's a better way to help people find, utilize, and synthesize information."
There's no denying that chatbots are eating away at Google's search dominance, leaving Google in a passive position.
Google is also fighting back.
According to foreign media, the "AI Overviews" launched at the I/O conference is the biggest change in Google's search service since its establishment 25 years ago.
Google highlighted that AI can understand and answer more than a dozen questions at the same time, marking a significant step forward in the AI-centric "search revolution."
Old companies vs Silicon Valley rookies, this showdown, there will be more good shows to watch.