• Starlink’s up-and-coming competition…
  • Google is sweating right now
  • A new generative AI makes its debut

Dear Reader,

Is it time for a “time out”? In need of a breather? Does the industry need to hit the pause button?

That’s exactly what a small group of people are demanding of the artificial intelligence (AI) industry. 

An organization called the Future of Life Institute has created a petition, and as of today collected 1,482 signatures for “all AI labs to immediately pause for at least six months the training of AI systems more powerful than GPT-4.”

The Future of Life Institute wants “independent experts” to use the pause to develop safety protocols for advanced AI design so that they are “safe beyond a reasonable doubt.”

Given the pace of development in artificial intelligence that we’ve seen in the last few months, the timing of something like this doesn’t come as a surprise.

The petition is written with a sense of urgency and the demand to pause development immediately. It reeks of fear that something the institute doesn’t want to happen will happen.

Some might interpret the petition as a representation of a fear that AI will become sentient and take over the world. Perhaps visions of Skynet come to mind. But actually, that’s not the institute’s concern at all.

What it wants is to control how these large language models (LLMs) are developed and to make sure that they are “safe,” “aligned,” “trustworthy,” and “loyal.”

And there’s the rub. It wants these generative AIs to be aligned with its desired narrative.

Its goal is to put safety rails in these generative AIs so that they present the “right” information that the institute wants us all to see. And it wants these generative AIs to be “loyal” to the desired narrative.

This petition is frightening, and I certainly won’t be signing it.

After all, who are they to determine what information we can and cannot see? 

And I don’t want an AI loyal to any group of “overlords.” That means it can be controlled. I want unbiased information from a generative AI, something that I can trust that the output hasn’t been manipulated.

And it is a stark reminder of the nonsense and violation of constitutional rights that we all experienced during the pandemic due to the policies employed by all the public health “experts” that we were supposed to trust with blind faith. As we’ve learned, everything they told us has been proven wrong.

As an interesting analog, the Great Barrington Declaration was created by some of the most well-respected and accomplished epidemiologists, virologists, and public health scientists in the world.

It proposed pandemic policies the opposite of what the “experts” were forcing upon the Western world. Over 60,000 virologists, epidemiologists, medical practitioners, and public health scientists signed the petition. And including concerned citizens, almost 1 million people have signed it.

That is a movement. And what they proposed was factual, evidence-based, and also based on peer-reviewed scientific research.

The “Pause Giant AI Experiments” petition is not. And with only 1,482 signatures, its opinions are not widely held in any way in the industry.

The reality is that the societal benefits of developing advanced AI as quickly as possible are immense.

This technology will put the most powerful AI in the hands of everyone in the world who has a smartphone. It’s the democratization of technology without any barriers other than access to a computer or a $100 phone.

The productivity improvements will be massive and material, and will happen faster than any previous time in history. And as we learned yesterday, it will increase economic activity, resulting in the growth of GDP (gross domestic product) by anywhere from 4–7%.

Just imagine the economic opportunity that this technology will create considering the world’s GDP is more than $100 trillion.

Forget the “pause”…full steam ahead.

Project Kuiper is gearing up for its commercial launch…

Big news from Amazon’s Project Kuiper: The team just made a major announcement that paves the way for a commercial launch.

As a reminder, Project Kuiper is Amazon’s response to SpaceX’s Starlink broadband internet satellite constellation. This will be Amazon’s own satellite internet service designed to compete directly with Starlink.

And the team just revealed the hardware that will serve as the receivers and transmitters for consumers. There will be three options.

We can think of these as satellite dishes… But they don’t look anything like the traditional satellite devices we are familiar with.

Here’s what will be the most common option designed for residential use:

Source: Amazon

Here’s the satellite receiver that will be installed on rooftops. We can see it’s quite futuristic looking.

And notice the cardinal in this image. This gives us a sense of just how compact Amazon’s satellite hardware is.

In fact, this particular device is about 11 inch square and just one inch thick. And it weighs just about one pound (minus the mounting bracket).

Amazon revealed that it can manufacture this hardware for just under $400. That’s a great price point for getting new customers up and running. 

Amazon could potentially sell the equipment for something like $249 to make it attractive and make up the rest on the monthly internet service fees.

And Amazon claims that this particular device can enable speeds up to 400 megabits per second (Mbps).

We don’t know if that’s true – Amazon hasn’t demonstrated it yet. But if it is, that’s faster than what Starlink has advertised. But just like with Starlink, broadband speeds will entirely depend on how many people are downloading off of a single satellite.

So Amazon is clearly trying to position Project Kuiper as a higher-performance alternative to Starlink. That certainly would help with marketing.

And remember, this compact rooftop receiver is just the first option. The second is a mobile antenna that can be used remotely, out on the road. Here it is:

Source: Amazon

As we can see, the mobile antenna isn’t much bigger than a cup of coffee. It’s perfect for traveling.

As long as consumers have access to a power source, they can plug in this antenna and get relatively fast internet connectivity from nearly anywhere – even if there’s no cellphone service. I love it.

Then there’s Project Kuiper’s enterprise offering. Here it is:

Source: Amazon

This is the device that Amazon will market to larger enterprise customers.

As we can see, it’s about the size of a standard solar panel. And Amazon claims that it will deliver speeds up to 1 gigabit per second (Gbps). That’s 5G-level performance.

Amazon’s plans are to launch two prototype satellites in the coming months for testing. And then it will begin mass-producing satellites by the end of this year.

That puts the company on track to begin offering commercial services to consumers around the second quarter of 2024.

It’s great to see Project Kuiper making such strong progress. We absolutely want to have competition in the satellite broadband internet space.

That said, Starlink began offering commercial services to consumers in October 2020. That means it’s going to have nearly a four-year head start by the time Project Kuiper goes live.

That’s a huge advantage. It means that Starlink will have basically mopped up just about all of the low-hanging fruit in terms of customer acquisition by then.

As long as Starlink continues to satisfy its customers, it will be very hard for Amazon to eat into its market share.

Plus, Starlink has a major head start on satellite launches as well.

Amazon’s going to need to launch thousands of satellites over the next several years just to be on equal footing. Starlink’s time to market in this case has created an incredible moat for its business.

I’m very glad to see Starlink has some competition coming. Competition always drives quality up and costs down.

That said, I expect Starlink to remain the front-runner with regards to satellite internet service for a long time to come.

Google is still struggling to catch up with OpenAI…

Speaking of competition – Google just made a move hoping to catch up on the generative artificial intelligence (AI) front.

As we discussed yesterday, Microsoft has leveraged OpenAI’s ChatGPT brilliantly. Microsoft has rolled out this incredible generative AI across its entire suite of products, including Microsoft Office and its Bing search engine.

And this has been a groundbreaking development in Microsoft’s attempt to compete with Google. In fact, it caught Google completely off-guard. This is the first time in the last 15 years or so since that’s happened.

In response, Google just started to roll out its own generative AI across Gmail and its Google Workspace applications. These include Google Docs, Sheets, and Slides – each direct competitors to Microsoft Office, Excel, and PowerPoint, respectively.

Google is also rolling out its new AI to Google Meet for video conferencing. This is a popular alternative to Zoom.

As for specifics, Google is calling its new generative AI PaLM. That stands for “Pathways Language Model.” It’s a large language model very similar to OpenAI’s ChatGPT.

And Google revealed that PaLM will be able to do things like draft, reply, and summarize emails in Gmail.

It will be able to proofread, write, and rewrite content in Google Docs. It will assist in putting together presentations in Google Slides. And it will help with formulas within Google Sheets.

All that sounds good. But there’s a nuance here.

Google has only made PaLM available to what it calls “trusted testers” in the U.S. Google hasn’t deployed this tech commercially. It hasn’t even made it available to all Google Workspace users.

This is a clear indication that Google is still very far behind when it comes to generative AI. And that’s a big problem.

As we’ve learned over the last few months, the pace of AI development is absolutely incredible right now. In the world of AI, months are the equivalent of years. That’s how fast this tech is advancing.

And that means Google is much farther behind than it may appear to be on the surface. Microsoft is rapidly capturing market share, as is OpenAI and soon Anthropic… And it’s going to be hard for Google to get it back.

Thinking bigger picture, there are dramatic implications here…

Over 3 billion people use Google Workspace products regularly. And it’s estimated that over 1.2 billion people use Microsoft Office products… And that’s probably a conservative estimate.

Put them together and over 4 billion people around the world are soon going to have access to the most powerful AI technology on the planet (Google’s PaLM and Microsoft’s OpenAI partnership combined). Essentially anyone with a computer or smartphone will have access to this incredible technology.

It’s like a light switch. Microsoft already turned it on. And Google will do so soon.

Never before in history has a disruptive technology been deployed this quickly at this scale. It’s hard to wrap our heads around just how transformational this is.

And here’s the thing – most people have no idea this is happening. And they certainly don’t know what’s coming.

We are going to see a dramatic transformation just in the coming months alone. By the end of this year, billions of people will be wielding super-intelligent and functional AI.

Obviously, that’s never happened before. And it’s hard to accurately predict what the world will look like when it does.

Anthropic’s Claude is now ready for prime time…

We’ll wrap up today with another big development on the generative AI front. Anthropic just announced its commercial release of Claude.

If you remember, Claude is Anthropic’s generative AI.

It’s designed to work a lot like ChatGPT. The only difference is that Anthropic is attempting to build a set of principles into Claude.

The team wants Claude to be grounded in three basic concepts: beneficence, nonmaleficence, and autonomy. They are calling it a “constitutional AI” for that reason.

And it certainly appears that Anthropic’s Claude is well ahead of Google’s PaLM right now.

We just discussed how privacy-centered search engine DuckDuckGo is experimenting with Claude. Anthropic is also working with Quora and Ocean Labs to deploy Claude across its products as well.

Claude is now being deployed through licensing deals in a similar way as ChatGPT. And Anthropic is focusing on enterprise applications first.

The team wants to apply Claude to things like human resource (HR) management, enterprise resource planning (ERP), cybersecurity, and workplace productivity and collaboration.

With this focus, Anthropic is angling to plug Claude into a wide range of software companies. Companies like Adobe, Oracle, SAP, Zoom, Palo Alto Networks, Fortinet, QuickBooks, and Zoom are among the big names that seem like a good fit.

As for its positioning, I’m sure Anthropic wants to position Claude as a more trustworthy AI than ChatGPT. Part of that comes from its focus on the three “constitutional” principles.

I suspect Anthropic will also suggest that Claude isn’t beholden to a large tech company in the same way as OpenAI and ChatGPT is to Microsoft.

That said, Google has now invested nearly $400 million into Anthropic.

It’s quite possible that Google is trying to do with Anthropic exactly what Microsoft did with OpenAI. That’s something we’ll want to keep a very close eye on. If Google’s own PaLM doesn’t perform very well, it could always fall back to Claude. 

That said, Anthropic noted that the size of the large language models that underpin generative AIs like ChatGPT and Claude have grown eight times a year every year since 2017. That’s extraordinary.

And it raises the obvious question – just how powerful will these AIs be at this time next year?

Again, we have never had a such a powerful technology advance and proliferate as fast as what we’re seeing with generative AI right now. Simply put, this technology is about to change everything.

So get ready. We’re in for one incredible ride…


Jeff Brown
Editor, The Bleeding Edge