Paul Stenhouse: Evolution in AI technology, skyrocketing valuations, and talent poaching

ChatGPT is ready to take on more involved tasks  

Agents are the next phase of the AI evolution, in which the AI can go away and can complete multistep tasks on its own. So, you could give it a bunch of information —PDFs, emails, spreadsheets— and ask it to find the narrative or insights and create a PowerPoint. It'll then go away, do it, and let you know when it's done. There's less hand holding and more 'reasoning' from the AI.   

It can also browse the web for you and take actions in the browser on your behalf –imagine a scenario where you could give it a photo of the contents of your fridge, give it your food preferences, tell it to come up with a meal plan, and even order the additional ingredients for you.  

 

Valuations are skyrocketing for these companies  

Bloomberg is reporting that OpenAI rival Anthropic is in the early stages of planning another investment round that could value the company at more than $100 billion.  

 

The mad scramble for talent is getting crazy  

Meta just poached two of Apple's key Artificial Intelligence leaders for its Superintelligence Lab. Two staff from Anthropic (Claude, the ChatGPT rival) left for another AI company, but quit after just a few days and returned to Anthropic.   

 

LISTEN ABOVE 

See omnystudio.com/listener for privacy information.

Episode Transcript

Speaker 1 (00:07):
You're listening to the Saturday Morning with Jack team podcast
from news Talks at be.

Speaker 2 (00:12):
Every week, AI seems to get better and chat gpt,
how now has a new function that extends beyond the
basic inputs when it comes to the large language models.
Our textbot, Paul Steinhouse is here to put it in
normal English for us. What does it mean?

Speaker 3 (00:26):
Paul? Yeah, maybe I should have just asked chat gpt
to do that for me. Shouldn't I please explain everything
you're actually doing in normal language. No, there's a new
word in the AI world, Jack that's taking storm, and
I think chat GPT is kind of leaning into this.
It's the agent, right. The agent is sort of the
next phase of this AI evolution, and it's where instead

(00:49):
of just you sitting there and typing something into the
thing and coming back with something, you can say do
a thing and it will go away and it will
actually take steps and think and be able to do
a whole series of tasks all on its own, and
multi step things too. And so what open aiyes now

(01:10):
release to these new agents on their paid plans. They're
coming to the free plan soon. But this is the
sort of thing that you're going to be able to
do right. So you could give it just a bunch
of information that could be just like PDFs or emails
or spreadsheets, all the kind of things, maybe from you know,
my documents folder at work. You throw it in there
and you say, okay, I need you to find the

(01:33):
narrative the insights across all of this information and credit
PowerPoint for me, right, and it will go away. It
will do it. It doesn't need the handholding. It does its reasoning,
peace and the magic. And we'll come back with that
type of thing. But the new one, the new agents
can even browse the web for you and even take
over a virtual computer and truly do things on your behalf.

(01:55):
So I was thinking this, would you know, this is
maybe not something you can do tomorrow, but see, we'll see.
You could take, say a photo of the contents of
your fridge. You could give it, you know, oh I
like this and I like that, give it some preferences,
and then you could tell it to come up with
a meal plan a bunch of recipes, and it can

(02:17):
even go away and put all that stuff into a
cart for you from your favorite supermarket.

Speaker 2 (02:21):
That is amazing.

Speaker 3 (02:22):
Yeah, like that is now. Now it may not be
perfect and it may not be polished. Yeah, but that
is now where we're at in terms of what is
possible with AI. It's not just hey, can you do
something fun for me? It's like, no, actually need can
you do my meals? Yeah?

Speaker 2 (02:39):
Yeah, And the point is, I suppose that the stuff
is just getting better and better. So even if it
isn't perfect and one hundred percent polished right now, it's
only going to get better and better. And the more
that people feed in information, the faster it will improve.

Speaker 1 (02:52):
Right.

Speaker 2 (02:52):
That's how these kind of that's the kind of exponential
nature of this AI progress. It's it's kind of yeah,
it's remarkable and as you will no doubt have expected, Paul,
valuations are absolutely skyrocketing for these AI companies.

Speaker 1 (03:08):
Yeah.

Speaker 3 (03:08):
Yeah, because people that the advancements are happening so fast,
like faster than any kind of technology we've kind of seen,
and people want to be part of it, I think.
And so Bloomberg is reporting that the big open AI
chat gpt raval called Anthropic, they're going to be doing
apparently some another investment round that could value the company
at one hundred billion US dollars.

Speaker 2 (03:32):
That's amazing. That is just absolutely ridiculous. What do you
think the chances are maybe we should spend more time
on this at some point. What do you think the
chances are that some of this AI hype is never
going to come to fruition? Like, can you put a
percentage number on it? Can you say, you know, maybe
thirty percent of the hype is going to live up

(03:52):
to you know, live up to expectations in the next
decade or so.

Speaker 3 (03:56):
Do you know, I mean it being completely autonomous and
taking a whole bunch of jobs. Maybe not, But there's
actually a pretty good percentage, right, I mean, I've talked
about some of the technology that we've seen come and
go over the years, like three DTV was one that
I just never understood and couldn't understand why people were
getting excited about that sort of thing, like that's never
going to happen, And then like the NFT whole thing

(04:18):
that was happening. This is just the symptom of an
economy trying to work out what to do with some
interesting technology. But this does feel different because, and I
say it feels different because we're not that far away,
like you can take a photo. Like all of those
things I mentioned about the recipes and doing the meal
plan is all possible today. That's not like science fiction.

(04:39):
That's real. It can you can take a photo and
it can interpret what the items are from your fridge
and put a list. It can take that list and
create recipes. You know. The bit that it's now doing
with the whole browser piece is really cool. Now how
accurate that is. We're going to have to put it
through its paces. But that's like the last link in
the chain. So if you imagine that's just like the

(05:00):
photo of your fridge, think about all the things you
do each day at work, think about all the undane
tasks that go on. It's coming for those. Yeah, and
that's where I think it's that's where I think that
maybe the hype is real.

Speaker 2 (05:15):
What do they call it Amara's law? Right? Amara's Law
concerns the effects of technology and says that we tend
to overestimate, like the effect of technology or the impacts
of technology in the short term, but then underestimate in
the long term. So, yeah, even if just a sliver
of the AI hype kind of holds true over the

(05:36):
long term, we're inf some profound changes. Thanks so much. Paul.
Paul Stenhouse is our text bit with us this morning.

Speaker 1 (05:42):
For more from Saturday Morning with Jack Tame, listen live
to news talks that'd be from nine am Saturday, or
follow the podcast on iHeartRadio