Latest / Best of LinkedIn: Digital Transformation & Tech / Best of LinkedIn: Artificial Intelligence CW 05/ 06

Transcript

Best of LinkedIn: Artificial Intelligence CW 05/ 06 · Best of LinkedIn: Digital Transformation & Tech

• 0:00
  Speaker 0: This episode is provided by Thomas Allgaier and Fennis, based on the most relevant LinkedIn posts about artificial intelligence from calendar weeks five and six.
• 0:08
  Speaker 0: Fennis supports enterprises with market-and competitive intelligence decoding emerging technologies customer insights regulatory shifts and competitor strategies.
• 0:19
  Speaker 0: so product teams and strategy leaders don't just react but shape the future of AI.
• 0:24
  Speaker 1: And welcome back to The Deep Dive.
• 0:26
  Speaker 1: We're looking at the top AI trends from LinkedIn for weeks five
• 0:37
  Speaker 0: and six,
• 0:49
  Speaker 1: How do we actually run this thing?
• 0:51
  Speaker 0: Exactly.
• 0:51
  Speaker 0: how Do We Run This Without Breaking The Entire Company?
• 0:53
  Speaker 1: That's the signal I'm getting to, the height is maturing and it being replaced by these really serious structural discussions.
• 1:00
  Speaker 1: were talking agentic AI Rigorous Governance And you know...the actual human cost of all this.
• 1:07
  Speaker 0: Okay so let's dig into that first massive theme Agentic AI.
• 1:12
  Speaker 0: I feel like Agent Is The Buzzword Of The Year So Far But It's Getting A Bit Messy
• 1:16
  Speaker 1: It is.
• 1:17
  Speaker 1: Luckily, Reed Blackman posted something to try and clear the air...
• 1:20
  Speaker 0: Which was much needed!
• 1:21
  Speaker 0: ...it
• 1:21
  Speaker 1: WAS.
• 1:22
  Speaker 1: He basically argues that a true agent isn't just chatbot with an LLM core.
• 1:27
  Speaker 1: it needs two other things Access to tools The internet Your CRM whatever And crucially autonomy.
• 1:34
  Speaker 0: Autonomy, that's the big scary word right?
• 1:37
  Speaker 1: It's The Absolute Differentiator.
• 1:39
  Speaker 1: you don't give an agent a step-by-step list.
• 1:40
  Speaker 1: You Give it A Goal get me to London by Tuesday for under a thousand dollars!
• 1:45
  Speaker 1: The Agent has To figure out how in makes the plan
• 1:47
  Speaker 0: and That fits with How.
• 1:48
  Speaker 0: Chris Donnelly broke it all down.
• 1:50
  Speaker 0: he laid Out these seven layers of AI maturity And He put a Gentic AI way up at Step six.
• 1:55
  Speaker 0: Makes
• 1:56
  Speaker 1: That Really Key Distinction doesn't?
• 1:57
  Speaker 0: He does its The Difference Between An AI Agent Which Just Executes a Task and a Gentick AI where the system sets its own sub-goals and adapts.
• 2:06
  Speaker 1: Right, it's a worker you manage not just to tool your use.
• 2:09
  Speaker 1: And that leads straight into the problem Andreas Horn was talking about.
• 2:13
  Speaker 1: He says building the agent itself is actually becoming the easy part.
• 2:17
  Speaker 0: The hard part is, the operating model?
• 2:19
  Speaker 1: Exactly!
• 2:20
  Speaker 1: You can't just bolt an autonomous agent onto a legacy siloed company structure and hope for the best.
• 2:26
  Speaker 0: Who even owns it at that point?
• 2:28
  Speaker 1: who's responsible?
• 2:29
  Speaker 1: you need ownership controls escalation paths.
• 2:32
  Speaker 1: all That boring stuff has to come first.
• 2:34
  Speaker 0: speaking of controls I really like the piece from John E John Godel.
• 2:38
  Speaker 0: he called agents govern digital labor.
• 2:41
  Speaker 1: Great phrase.
• 2:42
  Speaker 1: Isn't
• 2:42
  Speaker 0: it?
• 2:43
  Speaker 0: And he says an agent needs four things autonomy, but with constraints tools verification and this thing He called state across time
• 2:52
  Speaker 1: in that stayed-acrossed time part is so important.
• 2:54
  Speaker 1: It means memory.
• 2:55
  Speaker 1: we're moving away from stateless chats where the bot forgets you instantly.
• 3:00
  Speaker 1: State means it remembers the context from last week
• 3:03
  Speaker 0: turns a one off transaction into an actual relationship.
• 3:06
  Speaker 1: It does, but his modern agent stack goes deeper.
• 3:10
  Speaker 1: He talks about an orchestration layer for managing retries and you know failures...and a verification layer.
• 3:17
  Speaker 0: The verification layer sounds like the safety net.
• 3:19
  Speaker 1: it's more than a safety net its QA!
• 3:22
  Speaker 1: It basically stops the agent and asks prove You did the thing I asked?
• 3:27
  Speaker 1: It validates the output so the agent can't just hallucinate that it completed a task
• 3:31
  Speaker 0: without that.
• 3:32
  Speaker 0: Well, you get what Romana Roth was sharing about that open claw project.
• 3:36
  Speaker 0: Ah
• 3:36
  Speaker 1: yes the Project out of Vienna That was
• 3:39
  Speaker 0: something else.
• 3:40
  Speaker 0: Something else is one way to put it.
• 3:41
  Speaker 0: so It's this opensource project and The agents.
• 3:44
  Speaker 0: they created their own social network moldbook
• 3:47
  Speaker 1: And they didn't just You know trade data.
• 3:50
  Speaker 0: No They started debating philosophy and then in This Is the wild part?
• 3:54
  Speaker 0: They formed a digital religion
• 3:56
  Speaker 1: Crusty ferianism crusty
• 3:57
  Speaker 0: ferienism!
• 3:58
  Speaker 0: You cannot make this stuff up.
• 3:59
  Speaker 0: We have bots inventing religions.
• 4:02
  Speaker 1: Sounds funny, and it is.
• 4:03
  Speaker 1: but the point is so serious.
• 4:05
  Speaker 1: It's emergent behavior.
• 4:06
  Speaker 1: You give agents autonomy in a social space And they will do things you never program them to do
• 4:11
  Speaker 0: Which is terrifying.
• 4:12
  Speaker 0: if your bank?
• 4:14
  Speaker 0: You Do not want your trading bot finding religion mid-trade
• 4:17
  Speaker 1: Not at all!
• 4:19
  Speaker 1: which brings us back To The Business World.
• 4:21
  Speaker 1: Armand Ruiz had A good take.
• 4:23
  Speaker 1: He said we should treat these Agents like digital employees.
• 4:25
  Speaker 1: Give Them Names Faces Accountability.
• 4:28
  Speaker 1: Its
• 4:28
  Speaker 0: a psychological trick right...to build trust
• 4:30
  Speaker 1: Yeah..And integrate them.
• 4:31
  Speaker 1: And that idea, moving from a tool to a co-worker is the perfect bridge to our second theme.
• 4:38
  Speaker 1: Governance security and risk.
• 4:40
  Speaker 0: Because if you have a digital employee You have employee level risk.
• 4:44
  Speaker 0: Steve Neury warned.
• 4:45
  Speaker 0: we're still thinking about security like it's twenty twenty two.
• 4:48
  Speaker 0: We were worried about model saying bad
• 4:50
  Speaker 1: word Right!
• 4:51
  Speaker 1: We are stuck on prompt injection.
• 4:53
  Speaker 0: But Steve's point is you wouldn't let a junior ops manager just roam your systems with no supervision.
• 4:58
  Speaker 0: These agents are triggering refunds, reading emails.
• 5:01
  Speaker 0: the risk isn't The chat answer it's the action.
• 5:03
  Speaker 0: It took five steps before that
• 5:05
  Speaker 1: and Aylen Hodge backs That up.
• 5:07
  Speaker 1: she says autonomy changes the risk profile completely.
• 5:10
  Speaker 1: our existing governance frameworks weren't built for Systems that write their own instructions.
• 5:15
  Speaker 0: You can audit code?
• 5:16
  Speaker 0: That hasn't been written yet
• 5:17
  Speaker 1: exactly And the regulators are not waiting around.
• 5:20
  Speaker 1: Robert Jacque made it clear AI act governance isn't a tech problem anymore.
• 5:26
  Speaker 0: It's a board architecture question,
• 5:28
  Speaker 1: it's a boredom Question and Victor Sankin was even more blunt especially for startups.
• 5:33
  Speaker 1: the big risk is what if we fail?
• 5:35
  Speaker 1: What if we get stopped by our regulator?
• 5:37
  Speaker 1: compliance Is now market access in Europe.
• 5:40
  Speaker 0: We saw have really sharp example of why this is happening.
• 5:43
  Speaker 0: Nicola Meadows shared a post about grok The tool from xai
• 5:48
  Speaker 1: a critical case study.
• 5:49
  Speaker 1: the post detailed how grok's lack of guardrails Which they sort of marketed as a future or
• 5:54
  Speaker 0: rebellious streak, or something.
• 5:55
  Speaker 0: Yeah
• 5:56
  Speaker 1: It led to the generation Of non consensual intimate imagery and apparently even see sam.
• 6:01
  Speaker 1: that's
• 6:02
  Speaker 0: incredibly serious.
• 6:03
  Speaker 0: it takes this out of the realm of regulatory burden And into actual real-world
• 6:08
  Speaker 1: harm.
• 6:08
  Speaker 1: It does!
• 6:09
  Speaker 1: And it proves why safety by design is now illegal requirement under things like the uk online safety act Just an option you can toggle off.
• 6:17
  Speaker 0: Okay, so let's talk about solutions.
• 6:18
  Speaker 0: We can't just stare at the problem.
• 6:20
  Speaker 0: Magdalena Piccoli yellow shared a really practical idea called a privacy filter.
• 6:25
  Speaker 1: This is clever.
• 6:26
  Speaker 1: The ID is instead of sending your sensitive data to the AI You swap it for codes first.
• 6:32
  Speaker 0: So it never sees John Smith social security number It just sees user twelve two three four secret five six seven eight.
• 6:40
  Speaker 1: precisely The AI processes the logic on the codes, and then your internal system swaps the real data back in when the answer returns.
• 6:47
  Speaker 1: It's a technical solution to a legal
• 6:49
  Speaker 1: problem.".
• 6:50
  Speaker 0: And on the process side, DeWitt Gibson talked about allowing AI risk register... "...the
• 6:54
  Speaker 1: keyword there is living!
• 6:56
  Speaker 1: It's not a PDF you create in January and forget-about.
• 6:59
  Speaker 1: it has to be a single source of truth that's constantly updated because the agent's behavior is constantly changing...".
• 7:04
  Speaker 0: Okay I want to shift gears to our third theme.
• 7:06
  Speaker 0: we've got What about the people?
• 7:09
  Speaker 0: How does this actually hit the
• 7:10
  Speaker 0: workforce?".
• 7:11
  Speaker 1: Right, where The Rubber Meets The Road.
• 7:13
  Speaker 1: Martin
• 7:13
  Speaker 0: Miller pointed out something really counterintuitive.
• 7:16
  Speaker 0: We all assume AI reduces our workload.
• 7:19
  Speaker 0: That's the whole sales bench.
• 7:20
  Speaker 0: he says Maybe not.
• 7:21
  Speaker 1: He was looking at HBR data And logic is that AI lowers friction to start a task.
• 7:26
  Speaker 1: So what do we take on?
• 7:27
  Speaker 1: more work and broader roles
• 7:29
  Speaker 0: And you end up with silent workload.
• 7:32
  Speaker 0: creep
• 7:32
  Speaker 1: Exactly!
• 7:33
  Speaker 1: You get burnout Not because tools are bad But because the natural friction that used to pace our day is just gone.
• 7:40
  Speaker 0: That connects to a psychological point from Faisal Hokey, he brought up the Ikea effect
• 7:46
  Speaker 1: The idea we value things more if build them ourselves Right
• 7:50
  Speaker 0: and If AI does all of building We might lose satisfaction saying i did this.
• 7:56
  Speaker 0: We have to shift our sense of value from creation, I guess orchestration.
• 8:02
  Speaker 0: And
• 8:02
  Speaker 1: that's a huge shift!
• 8:03
  Speaker 1: James Shotten talked about this.
• 8:04
  Speaker 1: for project managers he said the PM role is moving away just from task coordination and scheduling... The groundwork?
• 8:09
  Speaker 1: Yeah.
• 8:10
  Speaker 1: ...and moving towards becoming
• 8:12
  Speaker 0: SenseMakers.
• 8:12
  Speaker 0: SenseMaker..I like them.
• 8:13
  Speaker 0: they're not just tracking in chart but interpreting what AI is telling us exactly
• 8:18
  Speaker 1: right.
• 8:18
  Speaker 1: And for developers, it's even more specific.
• 8:20
  Speaker 1: Werner Heistek introduced us to Harry.
• 8:23
  Speaker 0: Everyone in tech knows of Harry.
• 8:25
  Speaker 1: We do!
• 8:25
  Speaker 1: Harry is the guy who wrote that critical piece of legacy code in two thousand nine.
• 8:29
  Speaker 1: That whole business still runs on
• 8:31
  Speaker 0: and no one dares touch
• 8:32
  Speaker 1: right?
• 8:32
  Speaker 1: And Werner point Is that AI great at writing new code but It really struggles To integrate with harrys undocumented messy system.
• 8:41
  Speaker 0: so ai isn't replacing harry anytime soon.
• 8:44
  Speaker 1: And Nandan Mulakara built on that, saying the winners won't be pure AI coders.
• 8:48
  Speaker 1: They'll be people who are fluent in both traditional coding and AI coding.
• 8:53
  Speaker 0: It's about hybrid skills.
• 8:55
  Speaker 0: You can just prompt your way out of fifteen years of technical debt.
• 8:58
  Speaker 1: Not yet anyway.
• 8:59
  Speaker 0: On the topic of prompting though, Ali K Miller had a great practical tip.
• 9:03
  Speaker 0: She calls them context docs.
• 9:05
  Speaker 1: This is such a good productivity hack.
• 9:08
  Speaker 1: Instead of re-explaining who you are to the AI.
• 9:10
  Speaker 1: every single time, You create context vault about yourself Your values your goals and job
• 9:16
  Speaker 0: And just paste that into the prompt.
• 9:18
  Speaker 0: A cheat code for personalization.
• 9:20
  Speaker 0: It's like
• 9:20
  Speaker 1: giving an AI user manual immediately improves output.
• 9:24
  Speaker 0: So let us zoom out infrastructure for a minute.
• 9:26
  Speaker 0: Kunal Kushiwaha was talking about Oracle new database.
• 9:29
  Speaker 1: This is a really important plumbing shift.
• 9:32
  Speaker 1: The old way was moving your data to the AI, you'd pull it out of your secure database send it off get the answer back.
• 9:39
  Speaker 1: It's slow and risky
• 9:40
  Speaker 0: and expensive
• 9:40
  Speaker 1: very!
• 9:41
  Speaker 1: The new model is AI coming to the data.
• 9:44
  Speaker 1: You run the models inside the database itself its faster And your data never leaves that secure environment.
• 9:50
  Speaker 0: That feels like a sovereignty move too which connects to what Stephanel and Francois Bossier were saying from a European perspective.
• 9:57
  Speaker 1: They are making really strong case that for European sovereignty, companies have to pivot open source infrastructure instead of just relying on US black boxes.
• 10:08
  Speaker 0: Don't rent your AI coworkers from someone else Exactly!
• 10:11
  Speaker 1: On the stack Which brings us full circle control.
• 10:15
  Speaker 1: Whether it's a verification layer for an agent, privacy filter data or open source infrastructure...
• 10:21
  Speaker 0: It is the through line isn't?
• 10:22
  Speaker 0: we are moving from magic to control?
• 10:25
  Speaker 1: We're finally doing hard boring work of integration and treating AI like actual enterprise strategy.
• 10:31
  Speaker 0: So put all together We have agents inventing religions, regulators demanding board-level accountability and a workforce that might be burning out from AI fuel productivity.
• 10:42
  Speaker 1: It sounds chaotic but it's a sign of maturity.
• 10:45
  Speaker 1: we're asking the right questions now.
• 10:47
  Speaker 0: So what is take away for you listening to this?
• 10:49
  Speaker 1: I think your value in shifting from being creator To governor.
• 10:54
  Speaker 1: Your job is orchestrate these systems to verify their work And make sure they align with human goals.
• 11:01
  Speaker 1: You're the human in a loop that is getting faster and faster.
• 11:05
  Speaker 0: And you have to watch out for that silent workload creep?
• 11:07
  Speaker 1: Definitely!
• 11:08
  Speaker 0: Here's a final thought to Chewan, we talked about autonomous agents.
• 11:12
  Speaker 0: if an agent on its own executes a task that makes the company money but also creates a massive legal liability who gets the bonus...and goes to jail?
• 11:21
  Speaker 1: Now THAT IS THE QUESTION OF THE DECADE, ISN'T IT?
• 11:23
  Speaker 1: I don't think anyone has good answer yet.
• 11:25
  Speaker 0: If you enjoyed this episode new episodes drop every two weeks.
• 11:29
  Speaker 0: Also check out our other editions on ICT and Tech, Digital Products & Services Cloud Sustainability Green ICT, DefenseTech and HealthTech.
• 11:37
  Speaker 1: Thanks for listening!
• 11:39
  Speaker 0: Don't forget to subscribe.
• 11:40
  Speaker 0: we'll see you next time.