Latest / Connecting the Dots / Connecting the Dots - 6-12-2025

Transcript

Connecting the Dots - 6-12-2025 · Connecting the Dots

0:00
Today's tech news moves fast, but understanding
0:03
it shouldn't be hard. That's why Snarfle Solutions
0:06
Group brings you Connecting the Dots, where we
0:09
break down what's happening in tech and why it
0:11
matters to you. Now, let's dive in with your
0:14
hosts. Welcome to Connecting the Dots from Snarfle
0:17
Solutions Group, right here in Sacramento, California.
0:20
I'm Alex. And I'm Morgan. Hey, everyone. Glad
0:22
you could join us. We're here to cut through
0:24
the noise of tech news, connect the dots, and
0:27
figure out how all this stuff actually impacts
0:29
your world. Exactly. And connecting those dots
0:32
with you today, it is Thursday, June 12th, 2025.
0:36
Yeah. And if you're stepping outside here in
0:38
Sacramento, well, it's looking like a beautiful
0:41
day. Sunny skies. Clear forecast high is hitting
0:46
around 90 degrees Fahrenheit. Can't beat that.
0:50
All right, a quick check on the markets before
0:51
we dive in. From Thursday's close, the Dow Jones
0:54
finished up at 42 ,668 .77. OK. And the S &P
0:58
500, that closed at 6 ,012 .93. Got it. All right,
1:03
so let's get into the tech side of things. We've
1:04
been going through our internal Snarfle Tech
1:06
article report, sifting through what's making
1:09
waves. And we've pulled out three key stories
1:11
today that really caught our eye, not just for
1:14
the tech itself, but for what they could mean
1:16
down the line. Definitely. Where should we start?
1:18
Let's kick off with Amazon. This could actually
1:20
change what you see when you're browsing. They've
1:23
just rolled out their free AI -powered video
1:26
generator. for ads. Oh, right. I saw that. It's
1:29
for sellers. Yeah, exactly. It's now widely available
1:32
across the U .S. after what, about nine months
1:34
in beta testing. And they've apparently updated
1:37
their AI image generator, too. So what does this
1:39
thing actually do for the sellers? Well, it's
1:42
pretty clever, actually. It lets them take their
1:44
existing product photos, you know, the standard
1:47
static shots, the ones on the white backgrounds.
1:49
Yeah, those. And the AI turns them into short
1:53
like. eight -second video ads. Amazon's calling
1:56
them low -motion videos. Low -motion. Okay, so
1:59
like subtle movements. Right. It adds pans, maybe
2:02
some animated bits, just makes it more dynamic
2:04
than a still picture. Or if they already have
2:07
some video clips, the AI can sort of remix those
2:09
too. And the AI knows what it's looking at. How?
2:12
It's trained on, well, a mix of open data, but
2:16
also, crucially, the product information that's
2:18
already in the seller's Amazon ads account. So
2:21
it's tailored. Okay, that makes sense. Who's
2:24
this really aimed at though? Is it for the big
2:26
brands seems like the opposite actually? Amazon
2:29
saying is primarily for the small independent
2:32
businesses You know the ones that don't have
2:33
a huge marketing budget or like a dedicated video
2:36
team Okay, so it's about leveling the playing
2:39
field a bit that seems to be the idea Yeah, their
2:41
VP mentioned helping sellers do more with less
2:45
getting them beyond white backgrounds to show
2:48
products you know, actually being used, like,
2:51
in real life. So it can show, like, a watch on
2:53
someone's wrist or someone drinking from a mug,
2:56
that kind of thing? Exactly. Or add little animations,
2:59
like a figure adjusting the watch or a camera
3:01
panning across. And it's customizable, too. Customizable
3:04
how? Sellers can add text, which can also be
3:07
AI -generated, throw in their logo, mess with
3:09
the clip length, change the order. It's not just
3:11
a one click and done thing, apparently. Huh.
3:14
And how fast is it? Because video usually takes
3:16
time. Get this, they claim we can generate a
3:18
series of six different video options in five
3:21
minutes or less. Wow, okay. Five minutes for
3:24
six videos. That's really something. Is it actually
3:27
working? Are sellers using it? Seems like it.
3:29
They said adoption was good during the beta,
3:32
and about half the products that used the AI
3:34
video tool, they were being advertised for the
3:37
first time ever with video. So it is helping
3:39
smaller sellers get into video ads. Looks that
3:41
way. And the payoff seems decent too. Sponsored
3:45
brand campaigns that use video apparently see,
3:48
on average, a 30 % higher click -through rate
3:50
compared to the static ones. 30%, that's significant.
3:54
Yeah, Amazon's really pushing this as a, quote,
3:56
real, practical application of generative AI.
3:59
Something that's ready to scale up big time.
4:02
Okay, but, there's always a but, right? What
4:05
are the potential downsides here for us as consumers?
4:07
It just means more video ads, maybe. Probably
4:10
more varied ones, yeah. But you've raised a good
4:12
point, downsides. The big one is always guardrails.
4:15
Especially with AI creating content. Like making
4:17
sure it doesn't generate ads for fake stuff.
4:19
Counterfeits. Exactly. Amazon says these AI ads
4:22
go through the exact same moderation process
4:24
as everything else. Sounds good, but... But,
4:27
past reporting, I remember seeing things from
4:29
the Wall Street Journal, Washington Post a few
4:31
years back. Their automated systems sometimes
4:34
missed counterfeits, right? They did. Now, the
4:37
techs likely improve, they say, fewer visual
4:39
glitches, like, you know, weird extra fingers
4:42
on people in the demos. But apparently, even
4:45
in recent demos, some details were off, like
4:48
illegible text on a watch face. Stuff you might
4:51
not notice at a glance, but still. imperfections,
4:54
shows the moderation challenge is real. And it
4:57
feeds into that broader trend, doesn't it? Adobe,
4:59
Meta, they're all building tools to automate
5:02
creative work, especially for ads. Yeah, which
5:04
is great for access for small businesses, like
5:06
we said, but it definitely fuels those worries
5:09
about jobs, you know, graphic designers, video
5:11
editors. What happens as these tools get better
5:14
and faster? It's that ongoing tension, isn't
5:17
it? Empowerment versus disruption. OK, let's
5:19
shift gears. Completely different topic, different
5:20
continent. Over to the UK. Right. The UK data
5:23
bill just passed parliament. And the big news
5:25
is what's not in it. A really controversial amendment
5:28
about AI and copyright got dropped. Yeah, that
5:31
caused quite a stir. There was this month long
5:33
back and forth like legislative ping pong between
5:37
their House of Commons and House of Lords. And
5:39
it was all about this amendment pushed by the
5:42
Lords and, well, basically the entire UK creative
5:44
industry, right? Some huge names involved. Huge
5:47
names. Sir Elton John, Dua Lipa, Paul McCartney,
5:50
Ian McKellen. A lot of heavy hitters. And they
5:54
wanted one key thing. Transparency. Meaning?
5:58
They wanted a law forcing tech companies to declare
6:02
exactly which copyrighted materials, songs, books,
6:06
art, films they use to train their AI models.
6:08
Okay. And why was that so important to them?
6:10
What was the fear? The core fear was that AI
6:13
companies, especially big ones based overseas,
6:16
would just sort of vacuum up all this UK created
6:18
content without asking without paying. And use
6:20
it to train AI that could then replace human
6:22
creators. That was the worry. Job losses for
6:25
artists, musicians, writers. They called it potential
6:27
thievery on a high scale in an open letter. They
6:30
argued that protecting their creative assets
6:32
wasn't just fair. It would actually help the
6:34
UK. They believed transparency would lead to
6:36
a proper licensing market. Companies would pay
6:39
to use the data. generating revenue and supporting
6:42
UK creativity, which is a massive 120 billion
6:45
sector sector for them. And this isn't just theoretical,
6:49
right? Having companies like OpenAI and Meta
6:52
faced lawsuits over using copyrighted stuff without
6:54
licenses? Exactly. So the creatives had real
6:57
grounds for concern, but the UK government didn't
7:00
agree, or at least not with putting it in this
7:02
bill. Right. They refused the amendment. What
7:04
was their reasoning? They basically said, we're
7:06
already looking into this. They have an ongoing
7:08
consultation specifically about AI and copyright
7:12
and they plan a whole separate AI bill later
7:15
on. And the argument against the amendment from
7:18
the government or tech side. The counter argument
7:20
was that forcing this transparency now would
7:23
hurt the UK's own AI industry just as it's getting
7:26
started. That it would stifle innovation and
7:28
make the UK fall behind other countries globally.
7:31
Though the bill passed without the transparency
7:33
rule. The creative side lost this round. They
7:36
did. One of the lords leading the charge, Baroness
7:39
Kidron, called it a pyrrhic victory for the government.
7:42
Like, they won the battle, but maybe lost something
7:44
bigger. And the creative groups are saying they'll
7:46
keep fighting. Oh, definitely. They've vowed
7:48
to continue pushing for these kinds of protections.
7:50
This fight isn't over. Okay, so the AI copyright
7:53
part was the big headline fight, but this data
7:56
bill, it does other things too, right? Things
7:59
that might affect people more directly. Absolutely.
8:01
Beneath the AI drama, there's some significant
8:03
stuff in there. For example, new rules making
8:06
it easier for bereaved parents to access data
8:11
belonging to their deceased children. Okay, that
8:13
sounds important. Yeah, and simplifying how data
8:16
is shared within the NHS, their health service,
8:18
which could improve care, and even, get this,
8:21
plans for a 3D map of everything underground.
8:24
Like pipes and cables. Exactly, to help coordinate
8:27
roadworks better, which honestly sounds amazing
8:29
if it works. Yeah, less digging up the same road
8:32
three times. But there were other concerns too,
8:34
weren't there, about privacy? Yes. Big concerns
8:37
raised, especially about changes to GDPR rules
8:41
around automated decision -making. Critics argued
8:44
the bill allows AI to make decisions on its own
8:47
in more cases. Decisions like? Things like job
8:50
applications, maybe loan approvals, potentially
8:53
even policing decisions with fewer safeguards
8:55
than before, and making it harder for you to
8:58
challenge a decision made purely by an algorithm.
9:01
That sounds like something people should be aware
9:03
of, less human oversight. Potentially, yeah.
9:06
There were also worries about UK researchers
9:08
getting first dibs on UK data, like NHS health
9:11
data, before big US tech firms could access it.
9:14
And questions about the burden on small businesses
9:16
to comply with new data roles, and whether data
9:19
might be transferred overseas without strong
9:21
enough protections. So yeah, lots of bubbling
9:23
under the surface of this bill beyond just the
9:26
AI copyright fight. It really shows how tangled
9:28
data, AI, and everyday life are becoming. Okay,
9:32
let's connect one more dot for today. Moving
9:34
from policy to pure tech innovation and funding.
9:37
Right. Let's talk about Multiverse Computing.
9:39
They're a startup based in Spain, originally
9:41
known more for quantum computing stuff. But they
9:44
just raised a ton of money, right? A Series B
9:46
round. A huge round. $189 million, which is roughly
9:50
$215 million US. Wow. What tech got... Investors
9:55
that excited. Is it quantum? It's actually something
9:59
they call Compactive AI. It's described as quantum
10:01
-inspired, using complex math, tensor networks
10:05
that their CTO pioneered, but its application
10:08
is very much in today's AI world. Compactive
10:10
AI. Okay, what does it do? It compresses large
10:13
language models, you know, the massive AI models
10:15
behind things like ChatGQT and other advanced
10:17
AI. Really big, complex ones? Exactly. This tech
10:20
can apparently shrink them down by up to 95%.
10:24
95 % smaller. Whoa! Okay, 95 % smaller. But does
10:28
it still work? Do they lose accuracy? That's
10:30
the key claim. They say they can do this while
10:32
only losing maybe two three percent in accuracy
10:35
So huge size reduction minimal performance hit
10:38
why shrinking them so important to save disk
10:39
space It's way more than that those big LLMs
10:42
are well, they're beasts They need enormous computing
10:45
power cost a fortune to train and maybe even
10:47
more importantly They cost a lot and use tons
10:50
of energy just to run the inference cost, right?
10:52
when you actually use the AI. Exactly. Which
10:54
often means they can only run efficiently in
10:56
big data centers on specialized cloud hardware.
10:59
It's a massive barrier. So Compactive AI tackles
11:02
that cost barrier? Directly. By making the models
11:04
so much smaller, they become much faster to run,
11:07
supposedly 4x to 12x faster. And that speed,
11:11
plus the small size, dramatically cuts those
11:14
inference costs. They're talking 50 % to 80 %
11:17
reductions. 50 % to 80 % cheaper to run. That's
11:20
huge. It really is. It could unlock using these
11:22
powerful models in places where cost was prohibitive
11:25
before. And what does that mean practically for
11:28
you, for me, for everyday users? Well, this isn't
11:31
just about cheaper cloud AI. Big vision here
11:34
is making these powerful LLMs efficient enough
11:37
to run locally. Locally, like on my phone or
11:40
my laptop. Exactly. On your PC, your phone, maybe
11:43
built into your car systems, running on drones,
11:46
even on tiny computers like a Raspberry Pi. So
11:48
powerful AI without needing a constant internet
11:51
connection to a data center. That's the potential.
11:54
Imagine AI apps running faster because the processing
11:56
happens right there. Maybe more private, because
11:59
your data doesn't have to leave your device.
12:01
New kinds of applications possible in areas with
12:03
bad or no connectivity. Okay, I can see why investors
12:07
poured $215 million into that. That sounds like
12:10
a fundamental shift. It really could be. And
12:13
Multiverse isn't just starting out. They already
12:15
have about a hundred clients globally, including
12:17
big names like Bosch, Iberdrola, even the Bank
12:20
of Canada. Wow. They've got traction. Yeah, and
12:23
160 patents. This funding brings their total
12:27
raise to around a quarter billion dollars. They
12:29
say revenue is doubling each year, helped by
12:32
fees and a new partnership to host their API
12:35
on Amazon Web Services. So investors are betting
12:37
this becomes like a core piece of future AI infrastructure.
12:41
That's the language they're using, calling it
12:42
a potential foundational layer, a quantum leap
12:45
for getting AI actually deployed out in the world,
12:47
not just stuck in labs or data centers. Making
12:50
AI smarter, cheaper and greener too, presumably,
12:53
if it uses less energy. Exactly. That efficiency
12:56
push is seen as critical to unlock the next wave
12:59
of AI adoption across that huge $106 billion
13:03
AI inference market. It could really change how
13:06
you interact with AI in, well, pretty much everything
13:09
in the coming years. Okay, quite a journey today.
13:11
So just to quickly recap the dots we connected.
13:13
Let's do it. We started with Amazon's new AI
13:16
tool, basically helping small businesses jump
13:18
into video advertising much more easily. Right.
13:21
Then we jumped over to the UK, where their new
13:24
data bill passed without that controversial AI
13:26
copyright transparency clause the creative industries
13:29
were pushing for, though the fight continues.
13:31
And we finished with multiverse computing and
13:34
their massive funding round for compact AI tech
13:36
that dramatically shrinks those giant AI models.
13:39
Yeah, potentially bringing really powerful AI
13:42
capabilities directly onto your personal devices,
13:44
making it faster, cheaper, and maybe more private.
13:47
It really highlights how fast things are moving
13:50
from practical business tools to national policy
13:53
debates to fundamental tech breakthroughs. Absolutely.
13:56
It's constantly evolving, constantly changing
13:57
how we work, create, and just live. Well, that's
14:00
our time for Connecting the Dots today. Remember,
14:03
we are here to help here at Snarful, so reach
14:06
out on our website if you have any more questions.
14:08
Thanks for tuning in. We'll catch you next time
14:10
on Connecting the Dots. That's a wrap on today's
14:16
top stories. Remember, tech doesn't have to be
14:19
complicated and we're here to help connect the
14:21
dots. If you've got questions or want to learn
14:24
more, visit us at snarful .com. Thanks for listening
14:28
and we'll catch you next time.