This is How Google Will Crush ChatGPT!

Hello to all my awesome readers, and welcome the 100s of new ones that recently joined!

Today’s newsletter is a big one in terms of length and importance!

You want to read this one!

Without any more faff, let’s dive in…

Will Google rule the (AI) world again?

I have said countless times over the last ten months that we cannot discharge Google in the AI race despite fumbling and stumbling over themselves with their AI products.

It feels like everything they launch is overshadowed by OpenAI and ChatGPT.

They are not winning the sprint! However, AI is not a sprint. It is a marathon ultramarathon.

While OpenAI is in the lead, backed by Microsoft's deep pockets, we see the contours of Google’s “comeback” in AI.

Today, I will share four major efforts that could propel Google back as the king of tech.

Bard

Bard is Google’s ChatGPT competitor. But as I have repeatedly mentioned, my preference is ChatGPT with GTP-4. I use it daily, and it has made me a productivity beast.

And while I have not found my jam with Bard, others say they prefer Bard over ChatGPT or Claude 2.

But things might change very soon - also for me…

I cannot see it replacing ChatGPT for the task I use it for today, but I may expand my daily use of AI with Bard.

How?

Let’s have a look…

A few weeks ago, Google released the long-awaited “extensions”.

You may have heard about OpenAI releasing its plugins for ChatGPT to expand its capabilities. This has been moderately successful - though it is useful for certain things (like internet access).

However, Bard’s extensions are somewhat like ChatGPT's plugins but only access Google’s products and services.

Sounds minuscule?

It is not!

It is HUGE!

This means you can access your Gmail, Workspace (Docs, Sheets, etc.), Drive, Maps, YouTube, and Google Flights and Hotels from within Bard.

In addition, there is a function to double-check the Bard answers (press the G button), and you can collaborate with others in the chat/prompt window (you can do this with ChatGPT, too)

Double checking the answers and collaborating with others is all well and good - but WAIT A MINUTE, we can ACCESS all of our stuff on Google’s infrastructure from Bard???

In my businesses, we all use Google products, including this email you are now reading (well, for the newsletter part, I use Beehiiv, but the emails are hosted on Google infrastructure).

This is what I mean when I say that I may start using Bard daily.

In other words, not only can I use Bard and access my data and files, but I can also access Bard straight from inside a Google Doc or my Gmail inbox.

[For clarity: You can use Gmail for your business email/your domain. For example, my PracticalAI email [email protected] uses Gmail so everything looks and works the same as my personal Gmail [email protected]. If you haven’t used it, it is really neat and so much better than Outlook/Microsoft 😀]

Let’s look at this video, and especially the use case from around 1 minute:

How crazy is that travel example from about the one minute mark?

I also happen to own a few travel businesses, and while we operate in a different segment than what is used in this example from Google, I understand how this will be a big challenge for a lot of travel companies…

What other examples can we think of?

In a recent episode of one of my favorite podcasts, Marketing Against The Grain, the hosts talk about an interesting business use case: What if they used Bard to analyze which of their episodes are the most popular (on YouTube) and then the AI check their calendar what time they did those recordings, and then cross reference them to find if there is a link between the most popular shows and the recording times… For context, I can add that the two hosts are on different continents (USA and Europe), so they probably have some scheduling issues time might make a difference in the “performance” of the hosts.

Another use case is search inside the Google toolset. Search in Gmail is notoriously bad. Personally, I don’t understand how the best search engine in the world has such a sucky search inside their uber-popular email tool! With Bard integrated, the poor search experience will be a thing of the past! Yay!

Who here has “a million documents” on their Drive? I know I do… In addition, I have a ton of shit on Dropbox… (oh lord…). Having an AI inside my Drive/Google Workspace will bring me even more superpowers. Now, I don’t need to search or read through a bunch of old documents to find what I am after. Heck, I am sure I don’t even remember half of what is in there… Now I can simply ask Bard to find things for me, make summaries, or tell me what content I have hidden in there... CRAAAAZY!

Here is another interesting idea. What if you start “tagging” of all your documents? You don’t need to sort and structure your Drive anymore. Just put it all into the main folder and ask Bard to “Show me all data tagged as XYZ.” Chances are that is overkill - the AI can probably figure it out by itself.

Imagine that you are doing some sort of research or writing a newsletter (like this one) - which I write in Google Docs, by the way. I can now ask Bard to find me more information on a topic from within Google Docs. I can ask it to check on YouTube. Maybe I remember that AI YouTuber extraordinaire Matt Wolfe talked about the AI voice generator Air, and I want a summary of his thoughts to see if that is something I want to include in my newsletter. Bard will do that for me!

The opportunities and use cases are endless…

A side note: This move by Google goes a long way to explain Microsoft's investment in OpenAI, in my opinion. Microsoft does not have the LLM capabilities of Google, and in order to not be left in the digital stone age, I think Microsoft understood that they needed to get access to a powerful LLM to stand a fighting chance against Google. Since OpenAI was independent but in serious need of cash, Microsoft saw its opportunity and offered up its big fat checkbook. As I have spoken about earlier, what Google is now doing with Bard is what Microsoft wants to achieve with CoPilot. This time, however, Google beat them to it!

Google Assistant

I chose to put the Google Assistant as a separate point despite it would fit perfectly fine into the Bard section above.

So what is Google Assistant?

Google Assistant is a virtual assistant that is available on mobile and home automation devices.

Now the Google Assistant is Bard enabled, finally making it useful… (a bit of a bold statement, I know).

I think Google will not only “crush” OpenAI with their new AI advances, but they might also take a stab at Apple’s shitty Siri thingy…

In my opinion, Bard inside Google Assistant gives us a peek into what the future holdes with AI.

Here are some things the assistant can do:

• Control devices: Google Assistant can be used to control smart home devices, such as lights, thermostats, and speakers.

• Get information: Google Assistant can be used to get information about the weather, traffic, news, and other topics.

• Complete tasks: Google Assistant can be used to complete tasks such as setting alarms, sending texts, and making appointments.

• Create content: Google Assistant can be used to create content such as poems, code, scripts, and musical pieces.

• Entertain: Google Assistant can be used to entertain users by playing music, telling jokes, and reading stories.

Google Assistant is available on a wide range of devices, including iPhones and iPads… 😀

Gemini

Gemini is the LLM from Google DeepMind, hailed the GPT-4 killer.

We don’t have much information about Gemini after its announcement back in May 2023. However, there are various drips here and there, so here is what we know so far.

Gemini is built multimodal from the ground up, integrating text, images, and other data types. This could allow for more natural conversational abilities.

Gemini is utilizing Pathways, Google’s new AI infrastructure.

Not that I know what that means, but hey, cool name…

From what I can understand, Pathways allows for scaling and training on diverse datasets (many different data types, for the rest of us). The logical conclusion then would be the link between it being built multimodal from the ground up, and it being able to train from multi/diverse datasets. I suppose/I think other systems train on one type of data (ChatGPT on text, Midjourney on images, etc.).

Either way, multiple sources of training could mean that Gemini will become the largest LLM created to date.

Gemini is a series of models (AIs) working together (like ChatGPT/most other AIs). But what Gemini might have that the others don’t (currently) have is reinforcement learning and tree search giving Gemini reasoning and problem-solving skills.

Honestly, I don’t know what this means technically, as I sort of “assumed” we had this in GPT-4 and Claude 2. But the way this is talked about, it sounds like something new and different.

I know, I articulated that well…

Another interesting prediction is that Gemini may utilise memory and fact-checking against sources like Google Search.

This is where having Google’s ridiculous datasets over “all the internet” over the last 25 years comes in as a power play…

AND…

…don’t forget that Google owns YouTube! You may know it as a video platform (which is correct); it is also the world's second-largest search engine after Google!!

Yes, that’s correct; Google owns the world's two largest search engines!

And in case you do not know, each video on Youtube gets transcribed into text by their AIs already, meaning that all the content on Youtube is available to Google and Gemini in text format!

We know that around 300,000 hours of videos are uploaded to Youtube every day…

Mind blown yet?

I told you not to discard Google!

Google Pixel Phone

Lastly, I decided to add the new Google Pixel 8 to this mix.

Why is a phone a big deal in AI?

Personally, I couldn’t care less about the phone (Apple fanboy here). But inside we have:

• Magical Editor: Use AI to move objects, change facial expressions, edit the background, and resize objects in your photos.

• Audio Magic Eraser: Remove unwanted sounds from your videos, such as traffic noise or people talking in the background.

• AI spam filtering: AI helps you get 50% fewer spam calls on average.

I get spam calls daily, especially from UK numbers (Russian scammers), so I hope Apple gets their shit together and launch something similar soon!

But Thomas, what's the big deal?

What makes this phone worth talking about?

Well, it is not the phone per se. It is more the fact that Google’s mobile phone operating system, Android, is the largest in the world with about 71% market share…

Google has a history of releasing new features on Pixel devices first and then making them available to other Android devices in later.

Android’s market share is shrinking, and Google is losing market share to Apple.

Thus, I think it is highly likely that Android will get Bard and/or Gemini integrated at some point in the near future.

In other words, the phone itself isn’t particularly interesting to me, it is more that it is another peek into what is coming down the pike for AI, mobile phones, and our lives.

The lurkers…

While this issue focuses on Google and what they have up their sleeves, three other companies are lurking...

For those paying attention, Meta has been talking about launching their own LLM to compete with Open AI’s GPT. Meta, like Google, is sitting on tons of data. So, I suspect Meta could become a serious contender in the race for AI dominance very soon.

A more surprising entrant is Microsoft. What? Yes, Microsoft. Sure, they gave billions of dollars for a stake in OpenAI, and frankly, that was a smart move. However, it is costly for them to use the GPT models. Thus, rumors have it that they have started building their own LLM. And least not forget, Microsoft is also sitting on tons and tons of data from Search (Bing), Productivity tools (Office 365), Operating systems (Windows), Browser (Brave), Email (Hotmail/Outlook), Gaming (Xbox + games), and Azure (cloud infrastructure). I am not sure how much of that they can use for training, but they have a lot of data…

Elon Musk of Twitter/X cut off access to the platform for third parties a while ago because he does not want AI companies using “his data” to train their models. It is rumored that X will develop their own LLM based on Twitter data (and probably more). And while I am not sure if Twitter/X alone is a good enough data source, we have to remember that Elon Musk owns the world's largest AI company already - Tesla.

Did you think Tesla was a car company??

Meh, it is a data company!

For the Tesla fans out there, do you even know how much data is collected by your car?

Also, Elon owns NeuraLink (AI chip for your brain) and StarLink, satellites, and rocket ships…

That’s all I have for today!

I know it was a lot.

Let me know what you think about it all! Hit reply or tweet me @thomassorheim 

AI’ll be back!

- Thomas

PS! 
How does a robot eat guacamole?

.

.

.

With computer chips!

(Today’s dad joke courtesy of Claude 2)

This is not the end. This is where the fun begins!