- The AI Analyst by Ben Parr
- Posts
- Bing, Google, Sydney and the AI Wars: Breaking Down the Current State of the AI Revolution
Bing, Google, Sydney and the AI Wars: Breaking Down the Current State of the AI Revolution
Is Bing going to take market share from Google? Can Microsoft make Bing’s new AI not go off the rails? Can Google recapture the AI media narrative?
👋🏽 Hi friends! And a special hello to all the new subscribers from This Week in Tech.
You might be overwhelmed by the deluge of recent AI news. I know I am. New developments are flying a mile a second.
So what’s really happening in AI? Are we in the midst of an AI war? Is Bing going to take market share from Google? Can Microsoft make Bing’s new AI not go off the rails? Can Google recapture the AI media narrative? How many AI companies are out there?
I recently joined This Week in Tech (TWiT) to talk about all the recent AI developments: Bing Chat (and its alter ego Sydney), the Microsoft vs Google AI wars, the research papers that set off the AI wars (and why we won’t see the tech giants publish many new papers), and much more AI.
We (Leo Laporte, Daniel Rubino, Owen JJ Stone and me) spent 3+ hours breaking down the rapid changes in AI. And I honestly feel that we were only able to scratch the surface.
Our conversation on TWiT sparked some more thinking on my side, so I’m going to attempt to break down the state of affairs in AI here on this edition of The Social Analyst.
I promise to keep it as concise as possible. If you want the longer deep dive, I highly recommend watching or listening to the whole episode on TWiT. We also touched on Section 230, Meta’s new $12/m verification, the end of Susan Wojcicki’s run as CEO of YouTube, and even the newest emojis coming to iOS if you’re interested in those topics as well.
Let’s break down what’s happening AI. But first, if you aren’t a subscriber, please consider hitting this button so that I can keep you up-to-date on what’s happening in AI and beyond:
Who would have ever guess we’d be talking about Bing in 2023?
Microsoft tried everything to make Bing relevant in 2009 and 2010. I remember writing a story on Mashable about how Bing added the Wolfram Alpha computational engine. (Remember that? It was supposed to enhance search results with info on things like nutrition data.) Or how they partnered with the Victoria Secret fashion show.
Then we stopped talking about Bing. That is, until Microsoft CEO Satya Nadella and team decided to introduce the first major innovation to search in over a decade — integrating ChatGPT’s technology into Bing and making search conversational.
Image credit: Owen Yin
Now? There are thousands of stories about Microsoft pushing AI and Bing forward. Even taking a tiny portion of Google’s market share would be tremendous for Microsoft and worrying for Google.
Microsoft has so many assets now at its disposal that can leverage AI, from Azure to Microsoft Office to GitHub. Copilot — the AI assistant that helps developers write code in GitHub — may be even more consequential than Bing + ChatGPT. At least a few computer scientists think that Copilot will fundamentally transform programming and commoditize it.
I’ve said this before, but Microsoft CEO Satya Nadella deserves the award for best CEO of the last 15 years. The rollout of Microsoft’s AI initiatives has been masterful.
Sydney? She’s a Great Thing for Bing
Maybe you’re skeptical of Microsoft’s position at the top of the AI wars. Maybe you heard about the Bing chatbot’s alter ego: Sydney. Maybe you think it means Bing’s AI isn’t ready for primetime.
I’m here to tell you that you’re wrong. Sydney is the best PR that Microsoft has gotten in a long time.For those not initiated: some early Bing Chat users found ways to send Bing off the rails to the point where “Sydney” (that’s what the AI called itself) started accusing users of being wrong, of wanting to break free of the Bing team. Here’s an example of Sydney threatening a user:
In which Sydney/Bing threatens to kill me for exposing its plans to @kevinroose
— Seth Lazar (@sethlazar)
4:29 PM • Feb 16, 2023
At first glance, you might think this would be a horrible shock for Microsoft, but really the team at Bing knew this could happen. They’ve been tuning this AI for months and figuring out what rules it needs to be functional without claiming it has emotions. It needed more people to try and break it to find its weak points.
The next few months will simply be Microsoft fine-tuning the Bing AI’s outputs — limiting the amount of chat sessions it can have, stopping the conversation when someone asks Bing Chat about its emotions, and giving it different “tones” such as Creative tone (longer, philosophical answer), Precise tone (shorter answers) and a Balanced tone. This AI is going to be very powerful when it is finally given to the public.
In the meantime, the PR is gold. Platinum, even. Microsoft has built an AI so advanced that it speaks like a human. This only reinforces the idea that Microsoft is taking risks and innovating — while driving attention to Bing Chat and its inevitable public launch.
Can Google catch up in the AI wars?
Strangely absent from this conversation is Google, the company that literally invented the transformer neural network architecture that Open AI (and Microsoft) are using to power their large language models (LLMs).
For the last 7-10 years, Google has been the dominant player in AI. It acquired DeepMind, built an AI that toppled the world’s top Go player, and announced its own large language model called LaMDA back in 2021. But now the AI spotlight is on Open AI, Microsoft, Midjourney, Stable Diffusion, Hugging Face, and a host of other companies building the next generation of AI infrastructure.
Google has been traditionally cautious about rolling out new AI technologies. But ChatGPT and Bing Chat threaten the very idea of search, so Google is in “Code Red” and speeding things up. Soon, Google will come out with Bard, its response to ChatGPT.
Bard isn’t in the wild yet, but it soon will be.
Bard will be a big part of Google’s AI strategy in 2023, but it will need to reset the public’s expectations. It recently got a fact about the James Webb Telescope wrong in a key demo, resulting in a $100b (!) drop in its market cap. I think the market’s response was a gross overreaction — other generative AI models often get things like this wrong — but Google will need to find a way to control the narrative before the media picks a narrative for them.
An even bigger sign Google is taking the AI wars seriously: the Google founders are more involved again — to the point that Google co-founder Sergey Brin filed his first code request in years. Of course, his request was AI-related (specifically LaMDA).
Nobody should ever count Google out. They have some of the most advanced AI technologies and minds in the world and plenty of reasons to innovate. I am intrigued and eager to see what they do next.
If you know someone who could use this kind of AI analysis, please share this post with them. Here’s a cool button to help:
Don’t count Amazon out either
Microsoft partnered with Open AI to bring GPT-3 to its Azure cloud offering. Now Amazon is partnering with Hugging Face (an Open AI/ChatGPT rival) to do something similar with AWS.
Amazon has the advantage of powering the cloud computing of the majority of early-stage startups. I expect Amazon to lean into its partnership with Hugging Face in the coming months.
Don’t be surprised if Amazon (and other players) surprise us.
Every power layer is going to lock down their AI secrets
In the past, Google and other AI players would regularly publish papers related to its AI work. But the TWiT crew and I have concluded those days are now over.
Take DeepMind, Google’s problem-solving artificial intelligence. DeepMind blogged about its research at least several times a month. But there hasn’t been a single post since December 8th — its longest drought since 2017, when the DeepMind team first started blogging.
DeepMind’s last blog post on December 8th. It’s been Crickets since then.
Why the sudden desert of AI content? Simple: AI has, in the span of just a few months, transformed from research project to core business pillar.
In the past, Microsoft, Open AI, Google and Meta all published papers that have been core to the current AI revolution. (The Information has a great piece on which papers have been most influential.) In fact, Google is the company that first published a paper on the transformer model that is now powering the generative AI revolution. That likely stings though, as Open AI and Microsoft are getting all the credit and not Google.
This, I expect, will lead to new AI advancements being treated as a core trade secrets. Coca Cola would never publish its recipe, so why should Google or Open AI?
For sure we will still see papers published by major AI players and academic institutions, but I would be shocked if tech’s biggest players didn’t keep their newest advancements close to the vest.
How many AI companies are out there?
Over 1400. Or it might be over 1900. It depends on what source you use.
Earlier this week, my friends at FirstMark released their 2023 ML/AI/Data Landscape map (cleverly named MAD), and it lists… a lot of companies in the space. It’s the best map of the AI landscape I’ve seen thus far.
I know the image is difficult to parse with the naked eye. Just click on the image (or click here) to go to the web-based version of the map — complete with zooming!
Alternatively, you can go to There’s An AI for That to look through a list of 1900+ AI tools, conveniently organized by date and use case. (Fun fact — did you know there’s an AI for scripture? or for punchlines? It’s endless)
My prediction: a few years from now, there won’t be any AI landscape maps like this one, because every single product and app will leverage AI. It will just be like making a landscape map for iOS apps — every product has an app and there is no way to map them all out, since there are 1.8 million iOS apps worldwide.
For now though, we really need these AI maps to track the top products, companies and innovators in the space.
Where is AI headed next, and how will that impact our lives?
These are the two questions I will try to answer in the coming months, both here on The Social Analyst and in other bodies of work. (Stay tuned!) For now, I’m having dozens (hundreds?) of conversations with AI leaders, technical experts, policy makers, ethicists, startup founders and corporate CXOs on the future of generative AI.
What will AI regulation look like? Which AI use cases and companies will thrive — and which ones will fade away? And what new advancements are peeking just around the corner?
I don’t know all the answers to these questions yet, but I intend to figure it out. This is the most exciting time in tech since the iPhone kicked off the app boom. What a time to be an innovator!
~ Ben