- Artificial general intelligence is closer than we think…
- Need help with your e-mail? “Hire” Ellie the AI…
- A fourth new release from OpenAI…
Dear Reader,
It took just 36 days before this remarkable technology was banned.
It’s so powerful, so fast, and produces such good work that more than a million people used it only five days after it was released on December 1. And it didn’t take long for students to figure out that it was an incredibly fast way to complete their homework.
I am, of course, referring to OpenAI’s ChatGPT, which is a powerful artificial intelligence (AI) built on large language models (LLM) that make it effective in composing essays, writing screen plays, writing software code, and yes, even doing homework… all in a matter of seconds.
The adoption of ChatGPT spread so quickly that it led the New York City Department of Education to ban the use of the technology on January 5.
The Department of Education’s justification centered around the “negative impacts on student learning.” It’s now working on ways to identify if students are “cheating” through the use of this incredible technology.
This kind of policy position was inevitable. And it was amazing how quickly it happened. But it’s fundamentally wrong.
I certainly share concerns that the use of AI will result in the loss of critical thinking skills. But the reality is that the way of teaching these critical skills will need to adjust with the newfound presence of such incredible technology.
I recall both as a student, an executive, and an analyst spending countless hours and days pouring through voluminous amounts of reading material – much of which was of very little use in discovering what I was looking for.
The only lesson learned in such a time-consuming exercise was having the resolve to grind it out with determination to find the insights that I was looking for. While that’s worth something, it absolutely isn’t the best use of our time.
A tool like ChatGPT has the ability to eliminate those wasted hours and provide insights in a matter of minutes – even seconds – that may have taken days or longer.
Time is our greatest asset. It’s the one thing in life that we can’t control. There’s no pause button, no rewind, and no fast-forward to get us through moments in life.
We can’t take a breather, take a do-over, or speed through those painful moments. Time is relentless, persistent, and resolute in its purpose…
Which is why ChatGPT is so powerful.
It gives us time back almost immediately. It frees up time that would have otherwise been lost, or gives us time we would have otherwise not had to accomplish even more.
ChatGPT, and its equivalents, will become the smartphone moment for AI. Corporations and governments already using AI understand that power. Most consumers don’t even realize that they’re using AI every day.
But a personalized digital assistant that learns our patterns and preferences and quickly customizes itself to our individual needs will be something that everyone “must have.”
It won’t require any learning curve, as the interface will be voice or text. And it will make its users feel both empowered and important. It will be like a lightbulb – that moment when anyone using the technology immediately understands what the AI is capable of.
It’s happening right now. It’s that moment.
Interest in ChatGPT is currently so high, OpenAI had to limit access to it. It’s previously been free to try and use, and OpenAI hasn’t yet decided the business model for monetization.
This breakthrough is so significant, Google management issued a “code red” because the technology threatens its entire business. Yes, it’s that powerful.
OpenAI has yet to train ChatGPT for the purpose of internet searches, but just imagine if our searches weren’t manipulated, filtered, and in many cases paid for by the government to influence our thinking. Or by advertisers trying to lure us into buying their product or service.
Imagine what it would be like when we typed in exactly what we were looking for and the result came back every time, spot on, uncorrupted. Who wouldn’t want that? It’s easy to see why Google is running scared.
Hopefully, OpenAI won’t implement an advertising-based business model. If it does, we’re back to the same game that Google already plays.
A simply licensing model would suffice, perhaps set at $5 a month for a personalized digital assistant to radically improve our quality of life and make us more productive.
This isn’t a technology to ban; it’s a technology to celebrate.
It will lead to remarkable breakthroughs. And, if used properly, it will have the ability to provide students around the globe with world-class, unbiased education that’s tailored to each individual student’s best way of learning.
When I think about that, I get excited. It would completely democratize the kind of education that’s only been accessible to those who can afford private instruction. And that would be a remarkably good thing for the global society.
Artificial general intelligence is just a few steps away…
We’ve seen an absolute explosion of developments around AI over the last 12 months. It seems like we’re talking about remarkable new breakthroughs in these pages every week now.
And today’s no different…
Taking things one step further than the work that OpenAI is doing, a company called Keen Technologies has raised $20 million to focus exclusively on artificial general intelligence (AGI).
AGI has long been the subject of science fiction. And up until last year, most thought that it was more than a decade away.
AGI is when an AI becomes indistinguishable from a human. It can think, understand, learn, and apply its knowledge to solve any problem – just as we humans do.
And it has the world’s body of expert knowledge to draw upon. It will be like speaking to an expert in every field of study, all wrapped up in a single AI. Imagine having an accurate and specific answer to anything “knowable” in seconds for just about any question we can think of.
AGI is an area that’s been hotly debated in the tech industry. Some think that AGI is decades away. I disagree.
I predicted in 2019 that we’d see AGI by 2028. That’s just five years away. Some called me crazy, and another “expert” suggested that I clearly didn’t understand the technology.
I beg to differ. And given the developments of the last 12 months, I’m beginning to think that I may have been too conservative.
Thinking exponentially isn’t about having all the answers and knowing the exact steps that need to be taken for a technology to advance. It’s about being able to extrapolate the speed at which the technology develops and having the ability to see how advancements in other related technologies will enable the pace of development.
AGI is no longer a long shot. It’s quickly become a legitimate pursuit with a seeable timeframe to “creation.”
The CEO of Keen Technologies sees that as well. John Carmack is a well-known figure in the industry. He previously founded a gaming company called id Software. And he served as the Chief Technology Officer for Oculus, which is now Meta’s virtual reality (VR) division.
So Carmack has extensive experience in the tech industry and is an expert in how us humans interface with computing devices. And he clearly sees a path toward AGI.
Carmack is very much an optimist on the subject. He said that AGI will be like having a league of “universal remote workers” at our fingertips. AGI will help drive productivity in ways most of us cannot imagine today.
Of course, we’ll need to have protocols and procedures in place to ensure that we handle AGI in ethical and responsible ways. That’s going to be one of the great challenges of the next few decades.
But there’s no question that the opportunity is immense. I’m very excited to see what Carmack and his team come up with.
Keen Technologies is certainly a company to watch over the next couple of years.
The world’s most sophisticated email assistant…
Introducing Ellie – the world’s most sophisticated email assistant:
Ellie Can Write Our Emails
Source: Ellieai.com
As we can see, Ellie is an AI designed to help people manage their email. This is a great application.
Ellie is based on OpenAI’s GPT-3. And it integrates directly with the Chrome and Firefox web browsers.
From there, Ellie can process emails and assist users with drafting a response. Notice in the graphic above users tell the AI what kind of tone they want to use? They can also indicate whether or not they’re interested in the subject being discussed.
These inputs help Ellie draft a response on our behalf. Then we can make some tweaks and have a great email ready to send out.
What’s more, Ellie can analyze our own writing style and learn how to craft email responses that mimic the way we write and communicate. At that point, there’s an option to allow Ellie to respond to emails autonomously. That is, if we’re bold and trusting enough to turn the AI loose.
This is such a cool technology. And really, it’s just the precursor to having our own dedicated, intelligent, personal digital assistant. By that, I mean an AI that will be able to help with all our general tasks.
Ellie can already handle our emails autonomously. The next step is for an AI to be able to book appointments, make reservations, handle basic transactions, and otherwise help manage menial day-to-day tasks for us.
That’s where this is going. And I’m sure we’ll see more big breakthroughs in this area in the months to come.
In the meantime, if anyone wants to experiment with Ellie, we can try it out for free at https://tryellie.com/.
And as always, if any readers give the AI a whirl, I’d love to hear about it. Please share your experience with me right here.
OpenAI’s latest product release…
OpenAI is absolutely on fire right now.
First it was GPT-3. Then it was text-to-image AI DALL-E 2. Then ChatGPT. And now OpenAI has released a new AI called Point-E.
Point-E is a text-to-image AI similar to DALL-E 2. But there’s one big difference. It’s far more efficient.
DALL-E 2 is an incredibly powerful image generator. Here’s a great example of the AI’s work:
Point-E’s Text Prompt Imagery
Source: OpenAI
This image looks like it could be a photograph. But it was created from scratch by DALL-E based solely on human text prompts.
There’s a downside to this technology, however. It’s computationally intensive, as we would expect.
That’s where Point-E comes in. It can generate images using just a single graphics processing unit (GPU). This makes it far cheaper to use.
To do this, Point-E goes about image-creation that’s different from DALL-E. It takes human inputs and creates what’s called a “point cloud” based on them. Then the AI extrapolates and fills out the rest of the image from there.
This visual depicts the process:
Point-Cloud Imagery
Source: OpenAI
The two images on the right are point clouds that Point-E generated based on text inputs. They form the general shape based on a limited number of rough 3D “points” and color schemes of the desired image.
From there, the AI fills in all the details to create the final image. Those are the images on the left.
The advantage here is that these images require very little computational power to produce. That makes Point-E cheap and easy to manage. Of course, these images aren’t nearly as stunning as the ones DALL-E produces. They’re far more basic.
So it’s just a trade-off. For simple images, Point-E is the way to go. But for immersive, life-like images, DALL-E 2 is the right choice.
What I like about this is that we’ve gone from raw horsepower (DALL-E 2) to figuring out clever ways to make image-generation cheap and fast (Point-E). And OpenAI did this in just months.
The bottom line here is that OpenAI has hit its stride. This has quickly become one of the most prominent and valuable companies in the industry.
Just this week, OpenAI shares are rumored to be trading at a $29 billion valuation. And I’ve already predicted at least a $50 billion valuation before the end of this year.
Looking forward, we can expect to see GPT-4 released later this year. It’s already one of the most-anticipated tech events planned for 2023. And given what we’ve already seen from GPT-3 and ChatGPT, I’m certain that it will be mind-blowing…
Regards,
Jeff Brown
Editor, The Bleeding Edge