GPT-4, is said by some to be “next-level” and disruptive, but what will the reality be?

CEO Sam Altman answers questions about the GPT-4 and the future of AI.

Hints that GPT-4 Will Be Multimodal AI?

In a podcast interview (AI for the Next Era) from September 13, 2022, OpenAI CEO Sam Altman discussed the near future of AI technology.

Of particular interest is that he said that a multimodal model was in the near future.

Multimodal means the ability to function in multiple modes, such as text, images, and sounds.

OpenAI interacts with humans through text inputs. Whether it’s Dall-E or ChatGPT, it’s strictly a textual interaction.

An AI with multimodal capabilities can interact through speech. It can listen to commands and provide information or perform a task.

Altman offered these tantalizing details about what to expect soon:

“I think we’ll get multimodal models in not that much longer, and that’ll open up new things.

I think people are doing amazing work with agents that can use computers to do things for you, use programs and this idea of a language interface where you say a natural language – what you want in this kind of dialogue back and forth.

You can iterate and refine it, and the computer just does it for you.

You see some of this with DALL-E and CoPilot in very early ways.”

Altman didn’t specifically say that GPT-4 will be multimodal. But he did hint that it was coming within a short time frame.

Of particular interest is that he envisions multimodal AI as a platform for building new business models that aren’t possible today.

He compared multimodal AI to the mobile platform and how that opened opportunities for thousands of new ventures and jobs.

Altman said:

“…I think this is going to be a massive trend, and very large businesses will get built with this as the interface, and more generally [I think] that these very powerful models will be one of the genuine new technological platforms, which we haven’t really had since mobile.

And there’s always an explosion of new companies right after, so that’ll be cool.”

When asked about what the next stage of evolution was for AI, he responded with what he said were features that were a certainty.

“I think we will get true multimodal models working.

And so not just text and images but every modality you have in one model is able to easily fluidly move between things.”

AI Models That Self-Improve?

Something that isn’t talked about much is that AI researchers want to create an AI that can learn by itself.

This ability goes beyond spontaneously understanding how to do things like translate between languages.

The spontaneous ability to do things is called emergence. It’s when new abilities emerge from increasing the amount of training data.

But an AI that learns by itself is something else entirely that isn’t dependent on how huge the training data is.

What Altman described is an AI that actually learns and self-upgrades its abilities.

Furthermore, this kind of AI goes beyond the version paradigm that software traditionally follows, where a company releases version 3, version 3.5, and so on.

He envisions an AI model that is trained and then learns on its own, growing by itself into an improved version.

Altman didn’t indicate that GPT-4 will have this capability.

He just put this out there as something that they’re aiming for, apparently something that is within the realm of distinct possibility.

He explained an AI with the ability to self-learn:

“I think we will have models that continuously learn.

So right now, if you use GPT whatever, it’s stuck in the time that it was trained. And the more you use it, it doesn’t get any better and all of that.

I think we’ll get that changed.

So I’m very excited about all of that.”

It’s unclear if Altman was talking about Artificial General Intelligence (AGI), but it sort of sounds like it.

Altman recently debunked the idea that OpenAI has an AGI, which is quoted later in this article.

Altman was prompted by the interviewer to explain how all of the ideas he was talking about were actual targets and plausible scenarios and not just opinions of what he’d like OpenAI to do.

The interviewer asked:

“So one thing I think would be useful to share – because folks don’t realize that you’re actually making these strong predictions from a fairly critical point of view, not just ‘We can take that hill’…”

Altman explained that all of these things he’s talking about are predictions based on research that allows them to set a viable path forward to choose the next big project confidently.

He shared,

“We like to make predictions where we can be on the frontier, understand predictably what the scaling laws look like (or have already done the research) where we can say, ‘All right, this new thing is going to work and make predictions out of that way.’

And that’s how we try to run OpenAI, which is to do the next thing in front of us when we have high confidence and take 10% of the company to just totally go off and explore, which has led to huge wins.”

Can OpenAI Reach New Milestones With GPT-4?

One of the things necessary to drive OpenAI is money and massive amounts of computing resources.

Microsoft has already poured three billion dollars into OpenAI, and according to the New York Times, it is in talks to invest an additional $10 billion.

The New York Times reported that GPT-4 is expected to be released in the first quarter of 2023.

It was hinted that GPT-4 might have multimodal capabilities, quoting a venture capitalist Matt McIlwain who has knowledge of GPT-4.

The Times reported:

“OpenAI is working on an even more powerful system called GPT-4, which could be released as soon as this quarter, according to Mr. McIlwain and four other people with knowledge of the effort.

…Built using Microsoft’s huge network for computer data centers, the new chatbot could be a system much like ChatGPT that solely generates text. Or it could juggle images as well as text.

Some venture capitalists and Microsoft employees have already seen the service in action.

But OpenAI has not yet determined whether the new system will be released with capabilities involving images.”

The Money Follows OpenAI

While OpenAI hasn’t shared details with the public, it has been sharing details with the venture funding community.

It is currently in talks that would value the company as high as $29 billion.

That is a remarkable achievement because OpenAI is not currently earning significant revenue, and the current economic climate has forced the valuations of many technology companies to go down.

The Observer reported:

“Venture capital firms Thrive Capital and Founders Fund are among the investors interested in buying a total of $300 million worth of OpenAI shares, the Journal reported. The deal is structured as a tender offer, with the investors buying shares from existing shareholders, including employees.”

The high valuation of OpenAI can be seen as a validation for the future of the technology, and that future is currently GPT-4.

Sam Altman Answers Questions About GPT-4

Sam Altman was interviewed recently for the StrictlyVC program, where he confirms that OpenAI is working on a video model, which sounds incredible but could also lead to serious negative outcomes.

While the video part was not said to be a component of GPT-4, what was of interest and possibly related, is that Altman was emphatic that OpenAI would not release GPT-4 until they were assured that it was safe.

The relevant part of the interview occurs at the 4:37 minute mark:

The interviewer asked:

“Can you comment on whether GPT-4 is coming out in the first quarter, first half of the year?”

Sam Altman responded:

“It’ll come out at some point when we are like confident that we can do it safely and responsibly.

I think in general we are going to release technology much more slowly than people would like.

We’re going to sit on it much  longer than people would like.

And eventually people will be like happy with our approach to this.

But at the time I realized like people want the shiny toy and it’s frustrating and I totally get that.”

Twitter is abuzz with rumors that are difficult to confirm. One unconfirmed rumor is that it will have 100 trillion parameters (compared to GPT-3’s 175 billion parameters).

That rumor was debunked by Sam Altman in the StrictlyVC interview program, where he also said that OpenAI doesn’t have Artificial General Intelligence (AGI), which is the ability to learn anything that a human can.

Altman commented:

“I saw that on Twitter.  It’s complete b——t.

The GPT rumor mill is like a ridiculous thing.

…People are begging to be disappointed and they will be.

…We don’t have an actual AGI and I think that’s sort of what’s expected of us and you know, yeah… we’re going to disappoint those people. “

Many Rumors, Few Facts

The two facts about GPT-4 that are reliable are that OpenAI has been cryptic about GPT-4 to the point that the public knows virtually nothing, and the other is that OpenAI won’t release a product until it knows it is safe.

So at this point, it is difficult to say with certainty what GPT-4 will look like and what it will be capable of.

But a tweet by technology writer Robert Scoble claims that it will be next-level and a disruption.

Nevertheless, Sam Altman has cautioned not to set expectations too high.

More resources:


Featured Image: salarko/Shutterstock





Source link

By Rose Milev

I always want to learn something new. SEO is my passion.

Leave a Reply

Your email address will not be published. Required fields are marked *