The Inside Story of ChatGPT’s Astonishing Potential | Greg Brockman | TED

In a talk from the cutting edge of technology, OpenAI cofounder Greg Brockman explores the underlying design principles of ChatGPT and demos some mind-blowing, unreleased plug-ins for the chatbot that sent shockwaves across the world. After the talk, head of TED Chris Anderson joins Brockman to dig into the timeline of ChatGPT's development and get Brockman's take on the risks, raised by many in the tech industry and beyond, of releasing such a powerful tool into the world.

Follow TED!

#TED #TEDTalks #ChatGPT #ai

[chipkyle]

I'm 74 and am so glad to have lived to the see this beginning of AI. I carry in my pocket not only the Library of Congress, but also a tiny wizard to turn a world of facts into real-time useful knowledge. Had I lived in my grandparent's day, I would have been the first to take a ride in a horseless carriage. My Goodness, get aboard! It's going to be a wonderful ride. The challenge is not going to be this AI, but our cleverness in its use.

[Mrgasman]

Fasten your seatbelts, folks. As much as this guy puts it down in a funny, familiar, cute, and harmless way, this creation will make 2010 look like medieval times in a couple of years

[kushwahasiddharth]

It feels like I'm so lucky to be born in the late 90's and get to see the different eras of how our world is transforming into a whole new level of intelligence and innovations

[eduardobasto]

The pandora's box question might have a problem if we think of it as only a single box. If we think about it as 10 thousand boxes, spread around the world, each with a hundred people trying to open them, it might better represent the dynamic.


It will probably be a wild ride, but forfeiting the advantage of being among some of the first agents to open the box might just pass that role to agents whose intentions you don't know, just adding to the uncertainty.

[Learna_Hydralis]

How difficult to keep up with AI development is a strong indicator that life as we know it will never be the same!

[Youhydetube]

Greg, I stand in awe. Thanks for your amazing insights into the development of ChatGPT - the most incredible shift in information technology, research, writing and design since the Internet itself came into our lives.

[Neodynium.the_permanent_magnet]

Chris Anderson asked the right questions, some of which were evidently embarrassing, like "If there is a 1% chance of a Pandora thing in that beautiful box, do you open it?"
But the box has already been opened, as the outcome from GPT4 goes beyond expectations and intuition. And it's just the beginning.

OpenAI used to be "open". But then it's been proprietary closed software.
Thus another question: Should we continue to leave such responsibility in the hands of a few individuals within a private company?

[winkletter]

One thing Brockman mentions here that many people miss is the use of ChatGPT as an interface. While it's getting better at working on its own, it's really going to flourish as an interface between the end user and other applications. Soon we'll even see it using robotic process automation for applications that don't have APIs.

Another thing I'm excited about with these large language models are all the insights to be mined from all the data that was parsed. The models used by ChatGPT are a giant data source full of insights that can be queried with natural language. It can take unstructured data and reformat it as structured data. It can analyze and transform that data and, as Brockman demonstrated, search for insights in that data. We're going to start seeing questions answered that previously were too complex to explore.

[PaddySlattery]

I'm a quadriplegic who could be walking in a few years time because of this technological breakthrough, yet, at the same time, I yearn for the maturity of "those of you in control", at our peril. (I'm an optimist btw, and this post was written by me)

[Anurania]

One of the things that really impresses me about AI models is their personality. It's not like other tools that just complete the task, it offers explanations and feedback and helps to refine the output. That really helps to work with the computer. It feels as if it understands what you're doing and what you ultimately want. I've always tried to make my programs user friendly and intuitive, and AI takes that to another level.

[NaticzkaKaminskaHenryDolphin]

It's all incredibly complicated. First, the demonstration part about what's new about this machine was pretty dull. I understand that he's probably protecting the company's biggest secrets and some of the most advanced news, but what he showed is not that fascinating, at least for me. Secondly, although this technology certainly opens up a lot of new, fantastic possibilities, on the other hand, in addition to the obvious ethical and moral problems, there is also the problem of one small group lobbying, getting rich, and gaining all the power. As usual. His optimism may be 'cute' for some but I didn't chuckle when he said 'no one wants to read the whole book' nowadays. No, some of us actually do, and this is how humanity advances, still, and did in the past, including him. Writing and reading are the foundations of the whole process of learning.

A lot of people will lose their jobs because of this thing, and there will be even more depression in the world and suicide, so an even bigger mental health crisis than now - no doubt, yet we can't even resolve this current one.

But of course, neither he nor his colleagues and a very small group of various CEOs, rulers, politicians, and lobbyists will lose their jobs or power, nor money, on the contrary - they will only gain. Because even if AI would be able in the end to replace him/them, the creator, as all human experience, arts, etc predicted a long time ago, he still will have way more knowledge, power, and money to change it in time, before it happens, or just run away wherever, if it does. The rest of us, no matter the level of knowledge of AI that we will be able to gain from open sources, won't ever have the same resources as him.

Read about the bunkers of the super-rich that they all had built for themselves in the case of an apocalypse. It may sound ridiculous, like from a cheap, American, cliche action-catastrophe movie, but it's really true, and I have no faith in the good intentions and humanity of even the best ' most ethical' of these CEOs because I know how the world and our species work. All the empty talk about how respectful, full of love and goodness, and morality, you will be in the face of an existential threat is great until it actually happens. Generations of my family actually saw what 99, 99% of people do, when faced with unimaginable evil and impossible choices during World War II in Poland. And I'm not betting on this CEO and the current world leaders being that 0, 01% of heroic ones.

My view may sound cynical, dark, and bitter to some, but I just have no illusions anymore, I live too long in this world. All this talk about controls and guidelines, but it will end, as usual, that the obscenely rich will be even richer and the poor poorer. This trend continues in the 21st century, and this is actually a real problem.

And I'll warn the attackers right away: I'm not a fan of communism or an opponent of capitalism, hard work or success, not at all, quite the opposite, as a Pole, I saw with my own eyes, how communism ruins countries and societies (like any totalitarianism). However, I am opposed to the accumulation of insane fortunes at the expense of others, and in a dishonest way, which is absolutely what is happening now more than ever with many of these billionaire CEOs, only of course they prefer to show off cheap PR tricks to the public rather than have a real discussion and solve the problem.

As someone here suggested in the comments earlier, maybe all CEOs should be replaced by AI CEOs soon. But this also does not solve the problem at all, because behind all AI there will always be the same narrow, privileged group plus governments.
And we really need to talk about this particular part because AI will change the whole economic system forever.

[arthurfrancisco]

Im using Chat GPT4 as a tutor for coding bootcamp. Basically asking question and adding comments to my codes. If it still broken it will tell you what mistake syntax you've made. Don't use it to get answers but utilize this tool to ask question and reasoning. Using this tool to write the code for you will not work, but it does help with errors. Fact check and writing a perfect sentences are few things that does help cut time doing homework.

[iamthirdyt]

GDB!!! I remember seeing you walk out of Random Hall, dropping out of MIT to work at Stripe. I’ve always been astonished at your incredible intelligence, but also your attitude of humbleness and friendliness. Thanks for making humanity a better place.

[makstheimba]

We need more time to adjust and provide feedback.

There is no way humans can catch up with this pace of development. Right now I basically don’t even have enough time to watch summaries of all that happens in the field of AI, let alone understanding or adapting to change.

[HAZMOLZ]

I remember the birth of the internet and what a world changing technology that felt like at the time. That's taken about a generation to mature. I feel like each release of AI needs a generation to mature; but we get a month.

[tnijoo]

Kind of a lot of cop out answers at the end there, especially on the question releasing chatGPT too soon forcing competitors to unsafely rush just to avoid becoming obsolete.

[madhua]

This is the way to go. Release it early, soon enough to give time for the world to catch up and provide feedback rather then just one day unleashing it - which will no doubt astonish and freak people out. This is a good approach, well answered Greg.

[michaelperla]

He showed his true fear when he said “it’s always good to be polite” 😂😂 3:56

[PartridgeAves]

This is the first TED video I saw in which the audience is not showing their enthusiasm. They are terrified after every revelation this guy makes. Chris is also terrified, but this guy is like ‘Chill guys, nothing to worry about. It’s just a piece of cake.😢

[benjaminvandusen]

I have a lot of respect for the developers of this technology. These amazing people have worked their entire lives, studied, and obsessed over very complex astoundingly difficult technology to create. I am even more respectful that many are willing to suspend or halt the development of these technologies. Just the baby prototypes of these models alone are astoundingly useful, and enough to, hopefully, positively effect humanity.