Latest / Connecting the Dots / Connecting the Dots - 4-23-2025

Transcript

Connecting the Dots - 4-23-2025 · Connecting the Dots

• 0:00
  Today's tech news moves fast, but understanding
• 0:03
  it shouldn't be hard. That's why Snarful 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. It's Wednesday, April 23rd, 2025. Let's
• 0:20
  get you quickly up to speed. Weather first. If
• 0:23
  you're in the Southern or Central Plains, definitely
• 0:25
  be aware of some severe thunderstorms rolling
• 0:27
  through. You could see large hail, strong winds.
• 0:31
  Yeah, sounds pretty rough out there. And the
• 0:33
  South generally just lingering showers maybe
• 0:36
  for the next couple of days. Oh, and Western
• 0:38
  Hawaii, Kauai, Nihau flash flood watch from heavy
• 0:41
  rain. Wow. It's quite a mix across the country.
• 0:43
  But here in Sacramento, beautiful, sunny, high
• 0:46
  around 74, nice day. Can't complain about that.
• 0:50
  Markets wise, quick look back at yesterday, Wednesday
• 0:52
  the 23rd, Dow closed at 402 .51 .73 and the S
• 0:58
  &P 500 finished at five down four 49 .46. Okay.
• 1:02
  So today we're doing our deep dive. We've got
• 1:04
  three big tech stories and we really want to
• 1:06
  connect the dots and figure out how these things
• 1:08
  might. you know, actually touch your everyday
• 1:09
  life. Absolutely. These aren't just abstract
• 1:12
  tech news items. They have real potential impact
• 1:14
  on, well, how we find information, how AI develops,
• 1:17
  even how we connect with each other online. Right.
• 1:19
  So let's jump into the first one. It's that big
• 1:22
  U .S. government versus Google antitrust trial.
• 1:25
  It's been ongoing, but some some really interesting
• 1:28
  testimonies just came out. Yeah, the core issue
• 1:30
  hasn't changed, right? The DOJ thinks Google
• 1:32
  has an illegal monopoly. Search and search ads
• 1:35
  and a judge has already sided with them on monopoly
• 1:38
  point, which is huge It is but this new twist
• 1:41
  involves open AI exactly Nick Turley. He's head
• 1:44
  of product for chat GPT over at open AI He testified
• 1:48
  and what did he say get this? He said open AI
• 1:50
  would actually be interested in buying Google's
• 1:53
  Chrome browser buying Chrome Wow Yeah, if the
• 1:56
  antitrust folks force Google to sell it off the
• 1:58
  idea being it would restore some competition
• 2:00
  in search that would be Monumental. I mean, Chrome
• 2:03
  is the gateway to the web for so many people.
• 2:05
  Having an AI company like OpenAI potentially
• 2:08
  run it, that changes everything. Totally changes
• 2:11
  the game. Now Google, obviously they're not taking
• 2:13
  this lying down. They plan to appeal the monopoly
• 2:15
  ruling. And they're pointing fingers saying,
• 2:18
  hey, look at generative AI. Look at Meta. Look
• 2:20
  at Microsoft. There's plenty of competition.
• 2:22
  Right. They're trying to broaden the definition
• 2:24
  of the market. But here's another interesting
• 2:26
  bit from Turley's testimony. OpenAI actually
• 2:29
  tried to partner with Google not long ago. Partner
• 2:32
  how? They wanted to use Google's search tech
• 2:35
  inside ChatGPT. Apparently they were having problems
• 2:38
  with their provider at the time. Which turned
• 2:41
  out to be Bing, right? Microsoft Search. That's
• 2:44
  what we understand now, yeah. So they went to
• 2:46
  Google saying basically your search data is better,
• 2:49
  more up -to -date. Which really highlights how
• 2:51
  crucial that real -time massive data set is for
• 2:55
  these AI models. They need fresh, reliable info.
• 2:59
  For sure. OpenAI even said in an email something
• 3:01
  like, having multiple partners, especially getting
• 3:04
  Google's API access, would make ChatGPT way better
• 3:07
  for users. Makes sense. Diversify the data sources,
• 3:10
  improve the quality. But Google said no. Back
• 3:13
  in August, their reasoning... Concerns about
• 3:16
  like involving too many competitors keeping things
• 3:18
  close to the vest understandable from their perspective
• 3:21
  I guess but still pretty revealing definitely
• 3:23
  and it plays into what the DOJ wants as a potential
• 3:26
  fix, right? Charlie basically said if the DOJ
• 3:29
  makes Google share its search data with competitors
• 3:32
  like open AI Exactly. He said that would seriously
• 3:36
  speed up their efforts to improve chat GPT Give
• 3:39
  it that constant stream of current factual information
• 3:41
  it needs so That proposed remedy, the data sharing,
• 3:46
  it could be a massive boost for competitors.
• 3:49
  Seems like it. Turley even admitted chat GPT
• 3:51
  is still years away from handling most searches
• 3:54
  maybe 80 % with its own tech, so they really
• 3:57
  rely on external search data. So forcing Google
• 3:59
  to share could really level the playing field,
• 4:02
  at least in terms of beta access for AI development.
• 4:05
  That's the argument. And you know, the trial
• 4:07
  also dug into Google's past deals, like those
• 4:09
  exclusive agreements they had with Samsung, making
• 4:12
  Google search the default. Right, locking in
• 4:14
  that default position on millions of devices.
• 4:16
  They've apparently loosened up on those more
• 4:18
  recently. Samsung, Motorola, AT &T, Verizon can
• 4:22
  load other search options now. A little bit more
• 4:25
  choice then. A bit. But the DAJ wants more. They
• 4:28
  want to ban Google from making those big money
• 4:29
  payments just to get their search app installed
• 4:32
  by default in the first place. Ah, so cutting
• 4:35
  off the financial incentive to prioritize Google
• 4:37
  search. Exactly. So thinking about you, listening,
• 4:41
  how could this all shake out? I mean, if there's
• 4:44
  more competition forced here, could that actually
• 4:46
  lead to better search engines for you? More variety?
• 4:50
  Maybe better AI tools? It's a real possibility.
• 4:53
  More competition often drives innovation, better
• 4:56
  products, maybe even different approaches to
• 4:59
  privacy. It could change the fundamental way
• 5:01
  you find information online. Definitely something
• 5:03
  to keep an eye on. For sure. OK, let's shift
• 5:05
  gears. Story number two. NVIDIA. They've just
• 5:08
  announced the general availability of something
• 5:11
  called the NEMO platform. NEMO Microservices.
• 5:14
  And this is aimed at developers building. Well,
• 5:16
  they're calling them agentic AI systems agentic
• 5:19
  AI or AI teammates. They also said sounds very
• 5:23
  Futuristic it does but the core idea this data
• 5:26
  flywheel concept is actually pretty grounded.
• 5:28
  It's about continuous learning Okay, explain
• 5:30
  the data flywheel. So AI systems built with Nemo
• 5:33
  aren't static They're designed to constantly
• 5:36
  learn and improve based on how they're used they
• 5:40
  interact with enterprise data. They get user
• 5:42
  feedback And all that information gets fed back
• 5:45
  into the system to make the AI models better,
• 5:48
  smarter, more accurate over time. So the more
• 5:50
  you use it, the better it potentially gets. That's
• 5:53
  the core principle. It's a learning loop. Interesting.
• 5:55
  And NEMO isn't just one thing, right? It's like
• 5:57
  a suite of tools, curator, customizer. Exactly.
• 6:01
  Think of it as an AI development toolkit. NEMO
• 6:04
  curator handles the data. prepping it, cleaning
• 6:08
  it up, getting rid of duplicates, making sure
• 6:10
  it's ready for training. Huge task at scale.
• 6:13
  Right. Data prep is always huge. The Nemo Customizer
• 6:16
  is the training engine. It uses fancy techniques
• 6:18
  like fine -tuning to teach the models new skills
• 6:21
  or specific knowledge. OK. Nemo evaluators like
• 6:23
  the quality checker make sure the AI is actually
• 6:26
  improving and not forgetting things or getting
• 6:29
  worse at certain tasks. keeps track of progress.
• 6:32
  Important. You don't want AI regression. Definitely
• 6:34
  not. Then, NEMO guardrails. Super important for
• 6:37
  responsible AI. Keeps the AI on topic, makes
• 6:39
  sure it's safe, secure, avoids problematic content.
• 6:43
  Setting the boundaries. Precisely. And finally,
• 6:46
  NEMO Retriever helps the AI pull in the right
• 6:49
  information accurately, especially when lots
• 6:51
  of people are using it at once. reduces incorrect
• 6:54
  answers, handles scale. Wow. Okay, that sounds
• 6:57
  incredibly comprehensive. They seem to have thought
• 6:59
  through the whole life cycle. It seems that way,
• 7:01
  yeah. Now, we've heard about NVIDIA's NIMS before,
• 7:03
  NVIDIA Inference Microservices. How does NEMO
• 7:07
  relate to NIMS? Are they the same thing? Good
• 7:09
  question. Now, they're different but related.
• 7:11
  Think of NEMO as the factory or the workshop
• 7:14
  where you build and improve the AI model. Okay,
• 7:16
  the development side. Right. Data prep, training,
• 7:18
  evaluation, adding guardrails. Once you have
• 7:21
  that improved model from NEMO, Then you use NIMS
• 7:24
  to actually deploy it and run it. NIMS handle
• 7:27
  the inference part, the AI actually doing its
• 7:29
  job in the real world. Got it. So NIMO builds
• 7:32
  and refines, NIMS deploys and runs. And it's
• 7:35
  a cycle. Improve in NIMO, deploy via NIMS, get
• 7:38
  feedback, back to NIMO. Exactly. That's the flywheel
• 7:41
  in action, continuous improvement loop. Makes
• 7:43
  sense. And who's using this already? You mentioned
• 7:46
  developers, but... Are big companies jumping
• 7:48
  in? Yeah, they mentioned several. AT &T, Cisco,
• 7:51
  Galileo. What are they doing with it? Any examples?
• 7:54
  Well, AT &T is apparently using it to build an
• 7:56
  AI that can understand a huge amount of documents.
• 8:00
  Cisco and Galileo seem focused on using it for
• 8:03
  building better coding assistants for developers.
• 8:05
  So practical business applications right out
• 8:07
  of the gate? Seems so. And it's not just for
• 8:09
  NVIDIA's own AI models either. That's a key point.
• 8:12
  Oh. What else does it support? NEMO works with
• 8:15
  lots of different models, including popular open
• 8:17
  ones like Metis Llama, Google's Gemma, models
• 8:20
  from Mistral, Microsoft's Fi, plus NVIDIA's own,
• 8:24
  of course. So developers have flexibility. They're
• 8:26
  not locked into just NVIDIA tech. Right. And
• 8:29
  it integrates with other AI tools, too, like
• 8:31
  Lang chain and Lama index and data platforms
• 8:34
  from companies like HPE, Dell, IBM. It's designed
• 8:38
  to fit into existing ecosystems. OK, so bringing
• 8:41
  it back to the listener. How might this actually
• 8:44
  affect you? If you're not an AI developer at
• 8:46
  Cisco, why care about Nemo? Well, think about
• 8:49
  the apps and services you use every day. Down
• 8:51
  the line, many of them could have these AI teammates
• 8:54
  or agents built using platforms like Nemo integrated
• 8:58
  into them. Like a smarter Clippy, but actually
• 9:00
  useful? Potentially, yes. Or much more sophisticated
• 9:04
  assistants helping you at work. Maybe automating
• 9:06
  complex tasks, providing highly personalized
• 9:09
  information. And because of that continuous learning,
• 9:11
  the data flywheel, these agents wouldn't just
• 9:14
  be static tools. They adapt, learn your preferences,
• 9:17
  get better at helping you specifically. It could
• 9:20
  lead to much more capable and reliable AI assistance
• 9:23
  becoming part of your daily workflow or personal
• 9:25
  life. So smarter, more adaptive AI woven into
• 9:28
  more things we use could be quite transformative.
• 9:31
  It really could. Alright, final story for today.
• 9:33
  We're back in the courtroom this time with the
• 9:35
  FTC versus Meta. Right, another big antitrust
• 9:37
  case. This one focuses on Meta's acquisition
• 9:40
  of Instagram way back in 2012. ancient history
• 9:43
  in tech terms. But the FTC is arguing that buying
• 9:47
  Instagram wasn't just a normal acquisition. They're
• 9:50
  calling it part of a buyer burry strategy. Meaning
• 9:53
  Metta bought Instagram specifically to eliminate
• 9:56
  it as a potential competitor to protect Facebook's
• 9:59
  dominance in social media. That's the accusation.
• 10:02
  And just like the Google case, we got some pretty
• 10:05
  explosive testimony this week, this time from
• 10:07
  Instagram's co -founder, Kevin Systrom. Oh, interesting,
• 10:11
  what did Systrom say? He made a fortune from
• 10:12
  that deal, didn't he? He did, millions. But he
• 10:16
  testified that after Metta bought Instagram,
• 10:18
  the startup was, and this is his quote, starved
• 10:21
  of resources. Starved of resources, why? Because,
• 10:24
  according to Systrom, Mark Zuckerberg saw Instagram
• 10:27
  as a threat to Facebook's growth. Systrom literally
• 10:30
  said Mark was not investing in Instagram because
• 10:33
  he believed we were a threat to their growth.
• 10:35
  Wow. That directly contradicts what Zuckerberg
• 10:38
  testified just last week, doesn't it? Completely.
• 10:40
  Zuckerberg's line was that Metta poured resources
• 10:43
  into Instagram, helped it grow massively after
• 10:45
  buying it. So Systrom is basically saying, no,
• 10:48
  you deliberately held us back because you were
• 10:51
  scared of us competing with Facebook. That's
• 10:53
  exactly what it sounds like. It paints a picture
• 10:55
  of Metta intentionally kneecapping Instagram's
• 10:58
  potential to protect the mothership. Which, if
• 11:01
  true, is pretty much textbook anti -competitive
• 11:04
  behavior. It perfectly supports the FTC's buyer
• 11:06
  -bury argument. Right. It really lands a punch
• 11:09
  against Metta's defense. You have the co -founder,
• 11:12
  who benefited financially, saying the acquisition
• 11:14
  wasn't about synergy and growth, but about controlling
• 11:17
  a competitor. It makes you wonder, doesn't it?
• 11:20
  What would Instagram look like today if it had
• 11:22
  stayed independent? Or if Metta had actually
• 11:24
  poured resources into it without seeing it as
• 11:26
  a threat? How different would the whole social
• 11:29
  media landscape be? Exactly. And that's the relevance
• 11:32
  for you, the user. When big companies buy up
• 11:35
  promising startups, does it stifle innovation?
• 11:38
  Does it limit your choices in the long run? If
• 11:40
  potential competitors get absorbed or, as System
• 11:43
  alleges, starved of resources, you might end
• 11:46
  up with fewer platforms. Maybe less diversity
• 11:48
  in features, less pressure on the big players
• 11:50
  to innovate or treat users better. So as you're
• 11:53
  flicking through Instagram or Facebook, it's
• 11:55
  maybe well thinking about how these behind the
• 11:57
  scenes corporate strategies and legal battles
• 11:59
  actually shape the apps you use every single
• 12:03
  day. What alternatives might have existed? It's
• 12:06
  a good reminder that the platforms we use don't
• 12:08
  evolve in a vacuum. These acquisitions and the
• 12:11
  motives behind them have real consequences for
• 12:14
  competition and ultimately for us as users. Definitely
• 12:17
  food for thought. Okay, let's do a quick recap
• 12:19
  of those three stories. Sounds good. First, USB
• 12:22
  Google. Key testimony revealed OpenAI wanted
• 12:25
  to partner using Google search, but Google declined.
• 12:27
  The DOJ is pushing for remedies like forcing
• 12:30
  data sharing or even selling off Chrome. Big
• 12:32
  implications for search and AI competition. Second,
• 12:37
  NVIDIA's Nemo platform is now out. It's all about
• 12:39
  building these continuously learning AI teammates
• 12:41
  using that data flywheel concept. Supports various
• 12:45
  models. Big companies already using it. Could
• 12:47
  mean more adaptive AI in your future apps. Right.
• 12:50
  And third, FTC V Meta. Instagram co -founder
• 12:54
  Kevin Systrom testified Meta deliberately starved
• 12:58
  Instagram of resources post -acquisition because
• 13:01
  it was seen as a threat, directly contradicting
• 13:03
  Zuckerberg and supporting the FTC's case. Raises
• 13:06
  questions about acquisitions and innovation.
• 13:09
  You have three really distinct stories, but you
• 13:11
  can see the threads, can't you? scrutiny of big
• 13:13
  tech power, the incredible pace of AI development,
• 13:17
  and this constant tension between competition
• 13:20
  and consolidation in the industry. Absolutely.
• 13:23
  So the final thought for you today, as you see
• 13:25
  these legal fights play out, as AI gets smarter
• 13:27
  and more integrated, how do you think all this
• 13:30
  will change things? How you find info, how you
• 13:32
  use AI, how you connect with people online? What
• 13:34
  does that future look like to you? Lots to consider
• 13:36
  there. Definitely. We are here to help here at
• 13:38
  Snarfl, so reach out on our website if you have
• 13:40
  any more questions. That's a wrap on today's
• 13:45
  top stories. Remember, tech doesn't have to be
• 13:48
  complicated, and we're here to help connect the
• 13:50
  dots. If you've got questions or want to learn
• 13:53
  more, visit us at snarful .com. Thanks for listening,
• 13:57
  and we'll catch you next time.