Paul Stenhouse: The year of AI and the rise of passkeys

It has been the year of AI.. and it seems we're just getting started 

OpenAI is on track to hit $13B of 2025 revenue, up from $4B in 2024, according to The Information. It's looking at annualized revenue now of up to $19B. 

But, Merriam Webster has named 'slop' as the word of the year - the dangerous byproduct of AI use. 

Slop is "digital content of low quality that is produced usually in quantity by means of artificial intelligence". You start to see it on Reddit, emails, documents.. it's now just so easy to create 'text' that it can appear in abundance - which isn't always ideal.  
 

Passkeys are starting to have their moment too 

The new alternative to passwords which verify the website you're trying to log into before actually sending any information to them.

They're great because they eliminate phishing attacks, but.. they are a little tricky because unless they're shared to a password manager, they're stuck on that single device.

So if you can't access that device, or it's destroyed, stolen etc, then you can't login.

So, you need to make sure that 1) you sync them to a trusted manager like 1Password or a built in password manager like in Microsoft Edge and 2) that you have a recovery method, like a recovery email, set up on the account. 

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 b It.

Speaker 2 (00:13):
Has been a stonking year for AI. So just look
at open ai, the makers of chat GPT. They are
on track now to thirteen billion US dollars in revenue
for this year, which is up from four billion US
in twenty twenty four. But they're looking at maybe hitting
annualized revenue of about nineteen billion dollars, which is extraordinary.

(00:33):
That's just from the paid user accounts on chat gpt,
and of course you can just use the free version.
Texbert Paul Stenhouse is here to look back at the
year in artificial intelligence. Paul, did they get as far
as you expected they would get this year?

Speaker 3 (00:48):
Oh?

Speaker 4 (00:48):
Good question. I think they did.

Speaker 3 (00:51):
To be honest, Jack, I use a lot of I
say that as someone who uses AI a lot for
coding and technical things, and it seems to be one
of the areas where I've.

Speaker 4 (01:02):
Really leaned into it.

Speaker 3 (01:04):
There seems to be a war at the moment between
open ai and Google, Gemini and Anthropic to really kind
of nail the software developer experience. They're typically people who
want to try new things that are usually pretty.

Speaker 4 (01:18):
Okay with change.

Speaker 3 (01:20):
And what's interesting about code is there's a sort of
corpus of old projects you can reference and also kind
of documentation as well. Right, so it lends itself quite
well to AI. But it still isn't perfect.

Speaker 4 (01:32):
But when I think about.

Speaker 3 (01:34):
The things that it could largely could not do a
year ago or two years ago, Wow, it has come
a really really long way. It is not perfect, but
it's come a really really long way. I think it's
just interesting too that the people who probably have never touched,
of touched or thought of AI, maybe this year gave

(01:54):
it a go. I feel like it's reached a far
wider group of people. They've kind of started to dabble
as things have become, you know, a little more baked
into some of the services that they use every day.

Speaker 2 (02:06):
So what is your preferred text generation service at the moment?
Do you still stick with chat GPT, because in recent
months there has been a bit of a sense that
Google is catching up, that Gemini is actually getting as good,
if not maybe better than chat GPT.

Speaker 3 (02:25):
Yeah, And it really depends what you want to use
it for. And I know that sounds like a bit
of a cop out, but if, for example, in the
coding world.

Speaker 4 (02:32):
It seems to change.

Speaker 3 (02:33):
Every literally every few weeks, as these models get tweaked
and they're making changes and deployments. Now when it comes
to things like you know, some are slightly more creative,
some are better with academic some are better with more
formal type of writing, some are better at puzzing PDFs.
And I'm making understanding it really does change based on

(02:54):
the type of task that you're doing, which I know
sounds crazy, but they do sort of have this this
is going to sound even crazier. They almost have this
kind of like personality behind them jack or like a
skills because they are trained on things, and they do
have these kind of hidden instructions telling them how they
should be working and how they should be responding and

(03:15):
if they should be you know, responding more succinctly or
giving you more sources. And all of these things are
baked in the background, so things that you don't even
see about those agents are kind of controlling them. So
I would say try some of them. There's lots of
them that start out for free. I found things like
perplexity to be really good for kind of current from

(03:36):
trying to do anything about current information. I find you know,
the Geminis of the world. If you're a Google Drive
or Google kind of workspace user, it's really easy to
tap into that world and get it to understand some
things about the documents.

Speaker 4 (03:49):
You already have in there. So yeah, it's just you
can just you can kind of pick and choose.

Speaker 2 (03:54):
You can experiment, yeah, yeah, and try and different things.

Speaker 3 (03:58):
Yeah, and then music dedicated apps too, right, So things
like if you've ever made a PowerPoint presentation and you
never want to have to do that ever.

Speaker 4 (04:04):
Again, there's an app called Gamma, for example.

Speaker 3 (04:07):
So we're starting to see really specialize AI tools that
you can use to do very specific the Yeah.

Speaker 2 (04:14):
Right, okay, So can I put you on the spot
and ask you for a couple of predictions in twenty
twenty six do you think Okay, first of all, do
you think that we are going to achieve this agi
artificial general intelligence, which is the kind of you know,
the kind of guiding staff for many of these big
tech firms. They want to basically achieve artificial intelligence that

(04:35):
can do anything and more that a human can do.

Speaker 4 (04:38):
Not in twenty twenty six, Yeah, I don't.

Speaker 2 (04:41):
Think so either. I don't think so either. I think
if anything, even though the large language models, the text generators,
the chat GPTs of the world are improving, and certainly
that the models that are helping with coding have improved massively,
I think we're probably not. It feels like the kind
of incremental improvements now as opposed to really big lockstep improvements,

(05:03):
you know what I mean.

Speaker 3 (05:04):
Here's what's interesting, right, it can produce a lot of stuff.
And that's why the Merriam Webster word of the Year
is slop because that's what it's known as AI slop. Right,
These things just create stuff, whatever that is. But when
you start to look at like coding is a really
specific example and a really good example because it can
produce code and you can test to see if.

Speaker 4 (05:24):
It works very easily and very quickly.

Speaker 3 (05:27):
When it spits out sentences and sentences about quantum physics
or how to you know, look after a dog.

Speaker 4 (05:35):
You know, like, it will give you information. It will
give you words.

Speaker 3 (05:39):
They may be sentences, they may be paragraphs, but is
the content actually good and is actually correct?

Speaker 4 (05:46):
And it's sometimes when you.

Speaker 3 (05:47):
Really if you start to use AI for things you
know a lot about, or if you give it very
specific documents for example, and you're asking very specific questions,
it's very easy for you to then start to see
where it goes wrong.

Speaker 4 (06:01):
I heard someone.

Speaker 3 (06:01):
Refer to it the other day as a veneer of intelligence,
which I thought was quite interesting.

Speaker 2 (06:07):
Yeah. Nice, Okay, I like that, ive a here of intelligence.
My next My next question, repredictions. Do you think that
we are going to see an AI bubble burst.

Speaker 4 (06:18):
My portfolio? Short?

Speaker 3 (06:20):
Yeah, not yet. I don't think not yet. Okay, Look,
I think we still need a lot of things, right.
I just saw a news article today there's a lot
of pushback, for example, from locals on AI data centers
or data centers in general, that people don't want them
in their communities. They suck up utilities, drive up power prices,

(06:41):
et cetera, et cetera, kind of these types of things.
So we may, you know, we may see some really
interesting pushback that starts to kind of soften some of
this kind of AI hype. But look, isn't it the
art of isn't it the kind of a g I type,
you know, general intelligence level where it can really do
anything right, but it can do a lot of interesting stuff.

(07:01):
And I think where we will see is in this
next year when people really start to integrate AI into
their workflows, because it really does have kind of, you know,
in some fields, the intelligence quote unquote of like a
junior employee.

Speaker 4 (07:16):
When we start to see that, I think that's when
we'll see.

Speaker 3 (07:18):
What the real what the real impact is of the
economy and everything.

Speaker 4 (07:24):
Else, everything else.

Speaker 2 (07:25):
Okay, so it was a big year for AI. Next
year is set to be a big year for AI.
But you twenty twenty five was also a big year
for past keys.

Speaker 4 (07:33):
I know, Jack, it's the last show of the year.

Speaker 3 (07:35):
I usually always give my psa that says, if you've
got some time over the holidays, you should update your
past words. But twenty twenty five has been the year
of the past key, which it's this new kind of
alternative to passwords that's coming. And it's great because if
you think of a password, you kind of just yell
it at the website that you're going to, and if

(07:56):
you're the if it's the wrong website, you can hear
the password effectively. Right what past keys are? It goes, Hey,
who are you? And you go, I'm Jack, And I go, well, Jack,
I've got something for you.

Speaker 4 (08:06):
But only because you're Jack. And that's the past key.

Speaker 3 (08:10):
How is that the more secure they are, more they
are encrypted, they get stored significantly more securely on your device.

Speaker 4 (08:19):
You want to find out a little bit more about them.

Speaker 3 (08:22):
I do think the other way that everything is going.
They are a little tricky though, because they do kind
of get stuck on the device or the service at
which you generate them for right, So, because we need
to have that little handshaky thing. We do imagine if
you only have the paskey on your computer and then
you lose your computer. So you've got to put them
into things like password managers. You've got to make sure

(08:43):
you have things like a recovery email is set on
your account.

Speaker 4 (08:46):
Things like that. But they are great. They stop phishing.

Speaker 3 (08:51):
I don't see why every bank hasn't been adopting them because.

Speaker 4 (08:55):
They make you more seems like a god makes it
more secure.

Speaker 2 (08:59):
Hey, thank you so much, Paul, have a fantastic Christmas
and we will Catchiggy and very soon. Paul Stenhouse our textbook.

Speaker 1 (09:06):
More from Saturday Morning with Jack Tame. Listen live to
News Talks at B from nine am Saturday, or follow
the podcast on iHeartRadio.