Latest / Veramed Stories / Season 2 Episode 004 - The Human in the Loop: Matt Jones on AI at Veramed

Transcript

Season 2 Episode 004 - The Human in the Loop: Matt Jones on AI at Veramed · Veramed Stories

• 0:00
  Speaker 1: Hi everyone and welcome back to the VeraMed podcast.
• 0:03
  Speaker 1: I'm Lacey Wallace
• 0:04
  Speaker 2: And i am Rishali Goswami, today's co-host.
• 0:07
  Speaker 2: great to have you with us
• 0:10
  Speaker 1: Today episode is all about AI what it really means for our work Our industry The way we think of future.
• 0:19
  Speaker 1: We are
• 0:19
  Speaker 2: joined by someone who has been right at the center of that journey.
• 0:23
  Speaker 2: at VeraMed Mr Matt Jones.
• 0:26
  Speaker 1: We'll be exploring everything from practical use cases to big picture transformation.
• 0:32
  Speaker 2: And how we can embrace AI in a way that's responsible, human-focused and actually adds value?
• 0:40
  Speaker 1: Let's get started!
• 0:42
  Speaker 1: Welcome back to the Rebooted Podcast.
• 0:45
  Speaker 1: It has been minutes since Matt joined us last.
• 0:48
  Speaker 1: If anyone wants a time capsule moment.
• 0:50
  Speaker 1: His appearance was episode nineteen ten years of AirMed.
• 0:54
  Speaker 1: A lot has evolved since then, so for our new listeners here's a quick intro before we dive in.
• 0:59
  Speaker 1: Matt founded Varimet in twenty twelve and in twenty twenty five he shifted from CEO into EVP of data innovation at AI.
• 1:10
  Speaker 1: it is a move that reflects real curiosity and the willingness to explore how emerging tools can remove friction spark new ideas And support all of us on our day-to-day work.
• 1:21
  Speaker 1: He's been guiding our AI capability build with genuine care, making sure this journey feels responsible accessible and grounded in reality no matter where you are starting from.
• 1:32
  Speaker 1: So Matt welcome back.
• 1:34
  Speaker 1: let's jump right into the AI topic we're all here for.
• 1:37
  Speaker 1: Let's start on a personal note what motivated your desire to shift from CEO to EVP of data innovation an AI?
• 1:44
  Speaker 0: Well thanks for having me.
• 1:48
  Speaker 0: So, well I think after thirteen years as CEO that's quite a long time to do one specific role.
• 1:54
  Speaker 0: I saw AI starting to come with force into our industry and the world in general And my feelings is a result of that was a mix of excitement and fear probably unequal measure.
• 2:08
  Speaker 0: One of things.
• 2:09
  Speaker 0: our chairman always says this Do The Things That Only You Can Do?
• 2:13
  Speaker 0: an opportunity for me to take a role where I thought i could have the biggest impact both in protecting and enabling Veramed as a business.
• 2:22
  Speaker 1: Rashali, am gonna pass the baton to you now?
• 2:25
  Speaker 1: Yeah go for it!
• 2:27
  Speaker 2: Thank You Lissy.
• 2:27
  Speaker 2: um so Veramed has built its reputation on deep statistical expertise and trusted delivery from our sponsors.
• 2:36
  Speaker 2: As the industry is moving fast with AI new data models I would like to explore how you see that next chapter unfolding.
• 2:48
  Speaker 2: From ELT perspective, How do we see technology reshaping the CRO business model over the next three to five years?
• 2:57
  Speaker 2: And then in addition to that The Next Question will be...How Will We Wear Amid Or CROs Take Advantage Of That New World?
• 3:06
  Speaker 0: Yeah it's a great question and quite honestly i think were at start of very large shift in the use of technology for both CROs and all businesses worldwide.
• 3:20
  Speaker 0: In our industry, we've been long plugging away at standards, tech adoption tools... For the last twenty years I think that we have been trying to solve same problems in slightly different ways.
• 3:32
  Speaker 0: We achieved some things but I think efficiency and productivity hasn't necessarily kept pace with speed of adoption.
• 3:41
  Speaker 0: With AI coming into picture I can see the potential for a really big transformation.
• 3:46
  Speaker 0: Now, provided that as an industry we're able to not try to just do the same things that we are doing today with different technology if you could redesign what were doing and use AI to help kind of facilitate new way working because it's a very steady transformation.
• 4:05
  Speaker 0: so in your world but i think will take some time though now our industry they would know when on industries necessarily known for its speed and pace of change.
• 4:15
  Speaker 0: And yet this is a technology that's coming at us with the huge velocity, I think it will be a challenge we're going to need to overcome.
• 4:23
  Speaker 2: Okay sure thank you!
• 4:25
  Speaker 2: How do your balance AI innovation?
• 4:27
  Speaker 2: With non-negotiables quality and regulatory trust?
• 4:32
  Speaker 0: Well honestly there are no real compromises here.
• 4:35
  Speaker 0: my sense has been several conferences in last twelve or twenty four months.
• 4:40
  Speaker 0: Regulators might don't necessarily mind so much what tools we use.
• 4:45
  Speaker 0: We've been using SAS forever, where you started to use our Python and other languages for submission type work.
• 4:53
  Speaker 0: And the regulators I think will be guided by industry on What The Best Tools Are To Use.
• 4:59
  Speaker 0: That being said i Don't Think There's Really Any Compromise In Terms Of The Standards That We Will Be Held To As Companies Operating
• 5:07
  Speaker 1: in
• 5:08
  Speaker 0: really a high-stakes industry where we're dealing with patients, patient safety.
• 5:12
  Speaker 0: Where the drugs work and what their actions are over the longer term?
• 5:17
  Speaker 0: And so because of that I think will still be held to the same bar of accuracy reproducibility and traceability in everything we do.
• 5:26
  Speaker 0: there's been some regulations or kind position paper released this year at the start of twenty twenty six which you know essentially says AI is fine to use provided You can still prove what you've done, and it's right.
• 5:38
  Speaker 0: And its reproducible.
• 5:40
  Speaker 0: so I think It will take some time for AI to come but i don't the trust of What we do is going To change?
• 5:50
  Speaker 0: The issue that We got as an industry Is a whole Industry set up to be Human Accountable Everything we Do all the Regulations everything about human accountability.
• 6:01
  Speaker 0: So i Think we'll find it challenging to decouple Some Of the human layer from some of the technology layer and some of tasks that we do.
• 6:08
  Speaker 0: And because I think it will likely be a little bit slower than other industries, they aren't held to such high standards.
• 6:14
  Speaker 2: Thank you Matt for that assurance.
• 6:17
  Speaker 2: Now let's look at where AI actually has the promise To add value in clinical programming and biostatistics as well as automation Of standards and metadata
• 6:27
  Speaker 0: Have
• 6:28
  Speaker 2: come across realistic and high impact AI large language model use cases that involve emerging and ever-evolving biometrics programming tools.
• 6:40
  Speaker 2: And how much numeric, ontology based AI
• 6:43
  Speaker 0: models?
• 6:44
  Speaker 0: So I've seen a lot of promise from recent conferences i have been to.
• 6:50
  Speaker 0: it is moving very fast.
• 6:53
  Speaker 0: Lots of conference presentations which are about look at this magic what AI can do.
• 7:00
  Speaker 0: the narrative has started you know, not just what it can do but Can we actually stand behind?
• 7:06
  Speaker 0: What's being done?
• 7:08
  Speaker 0: and they're two very different questions because I think AI is definitely capable that the model advances.
• 7:15
  Speaker 0: That are coming through daily or weekly of proving out The power of LLMs when harness in an appropriate way But for our industry In particular where risk Is a big issue Where?
• 7:28
  Speaker 0: Transibility as a big issues a big issue around trust and being able to stand behind the results that we're producing or the code they were producing.
• 7:36
  Speaker 0: So I've seen a lot of promise for LLMs.
• 7:38
  Speaker 0: in terms like most recent, i think most companies out there are still talking about their possibilities with what AI can do as opposed you having delivered measurable value in terms of AI implementation in hard sense.
• 7:52
  Speaker 0: but where I'm seeing more success is when LLM's coupled knowledge context and data models that represent the underlying tasks explicitly.
• 8:01
  Speaker 0: And I think you asked about kind of numeric model non-tology based models, and numeric models.
• 8:06
  Speaker 0: from my understanding of those is essentially how an LLM works.
• 8:10
  Speaker 0: The ontology base models are where we've got more concepts or standards infrastructure that sits behind it.
• 8:16
  Speaker 0: That's when i have seen most success because the traceability assurance ability to keep LLMs on rails and create reproducible results is much, much higher than standard LLMs.
• 8:32
  Speaker 0: Now of course LLM's are trained on publicly available data um...and maybe some other information from subject matter experts and things like that.
• 8:41
  Speaker 0: but I think there's no getting away from the fact it was a huge tacit layer of knowledge that exists within organizations with an expert's mind that sits outside of LLMS today.
• 8:52
  Speaker 0: so I don't think here's a case in our industry specifically of saying, i'm just going to use an LLM.
• 8:56
  Speaker 0: To do this specific task because actually it is missing the context and knowledge that exists within specialist minds within our industry.
• 9:05
  Speaker 0: So where does the future lie?
• 9:07
  Speaker 0: The future lies in terms of equipping LLMs with standards And ontology based definitions.
• 9:15
  Speaker 0: Is equipping them with appropriate contexts.
• 9:18
  Speaker 0: It's equipping then with appropriate SME knowledge.
• 9:20
  Speaker 0: When you've got all those things harness together, that's when you get the real
• 9:24
  Speaker 2: power.
• 9:24
  Speaker 2: Very insightful thank-you.
• 9:27
  Speaker 2: can you describe some safeguards we have?
• 9:30
  Speaker 2: That ensure AI outputs are inspection ready and GXP and CD standards compliant?
• 9:37
  Speaker 0: yeah I don't.
• 9:38
  Speaker 0: again to the last point.
• 9:39
  Speaker 0: i didn't think there was any real difference from how we operate today.
• 9:43
  Speaker 0: We're not about to hand over the keys put the data in one end, get some results out at the other end submit those and hope for the best.
• 9:51
  Speaker 0: That's never going to happen in our industry a least not any time soon.
• 9:56
  Speaker 0: so For the time being I think we're still gonna need very similar And appropriate artifacts In place To prove what we've done and to prove why What We've Done is right.
• 10:06
  Speaker 0: Because of that i think will largely be seeing AI used in assistive use cases Uh...to supplement what humans are doing rather than necessarily displacing human skill sets.
• 10:16
  Speaker 0: By doing that we kind of get around the accountability issues.
• 10:20
  Speaker 0: In the future, I think this might change.
• 10:22
  Speaker 0: We're talking a lot about human in-the-loop as a concept and everyone feels very reassured by talking about the concept to Human in The Loop because it's great isn't if humans check what is being done?
• 10:32
  Speaker 0: But i know myself that when you can produce stuff using AI in seconds That would have taken you hours before And suddenly got fifteen or twenty pages information which looks great.
• 10:43
  Speaker 0: You know its quite difficult.
• 10:45
  Speaker 0: then go through and be accountable for what's being produced.
• 10:50
  Speaker 0: And I think in a similar way to how we prove things in our industry, in terms of QC&QA today by documenting who has checked what?
• 10:59
  Speaker 0: What they've checked... We'll likely see those kinds of processes evolve and get engineered into AI-enabled processes in the future.
• 11:08
  Speaker 0: If i'm a regulator and sitting there knowing that AI is used as part what human in the loop evidence is there to prove that everything has been checked accordingly.
• 11:19
  Speaker 0: So, you know I think AI as a potential game changer for all of these things and the GXP context on CDIS context.
• 11:25
  Speaker 0: but i think there are some challenges.
• 11:27
  Speaker 0: they're we need to get around to prove it can remain accountable for whats being done.
• 11:32
  Speaker 2: Thank You so much bringing our attention to this Human in Loop topic.
• 11:38
  Speaker 2: thats very important Thank you.
• 11:40
  Speaker 2: Is AI a productivity accelerator or is it fundamental redesign of how programming teams function?
• 11:47
  Speaker 0: A great question and the conference I was at this week, This was one of the main topics that was there.
• 11:53
  Speaker 0: There's kind of a big debate out there about whether we're building something brand new New processes, news systems, UK capabilities Or Whether this is a productivity accelerator Making us faster cheaper The things that we do today.
• 12:07
  Speaker 0: That latter topic is a bit of a taboo topic.
• 12:09
  Speaker 0: So just being faster and more productive, it's not necessarily been seen as the game changer.
• 12:15
  Speaker 0: that being said I think its where we're likely to see change first than what was like to see measurable value.
• 12:22
  Speaker 0: As i've said I don't hear the artifacts produced in terms of STTMs atoms tables figures listings clean lot databases submission packages etc.
• 12:33
  Speaker 0: are suddenly going to disappear overnight.
• 12:35
  Speaker 0: We have long regulatory cycles, long study cycles.
• 12:39
  Speaker 0: The speed at which change will percolate through our industry means that those assets will be around for a number of years to come.
• 12:46
  Speaker 0: so while that happens I think the AI will be a productivity accelerator.
• 12:52
  Speaker 0: That being said on top of that i think that AI gives us new capabilities but don't already exist today.
• 12:59
  Speaker 0: and AI It doesn't eat or sleep.
• 13:01
  Speaker 0: I mean, it eats tokens and tokens are becoming more expensive.
• 13:05
  Speaker 0: but its got the ability to crunch through vast amounts of information and data And gives us a capability that didn't previously exist.
• 13:13
  Speaker 0: When you think about pharmaceutical companies The back volumes of studies and data protocols analysis plans specifications etc.
• 13:24
  Speaker 0: There's huge goldmine there of legacy data assets code bases and information.
• 13:29
  Speaker 0: And there's potential to unlock significant value by getting insights from that data, the people can't already get today.
• 13:36
  Speaker 0: looking into the future I've got you know just one more point points of make on that.
• 13:39
  Speaker 0: i can see a world where we start to operate differently.
• 13:43
  Speaker 0: um i think they'll be new roles in our industry.
• 13:46
  Speaker 0: i think will think about workflows and data differently developing agents and workflows, I think we'll be having people who are managing knowledge and skills.
• 13:57
  Speaker 0: And i can also see our standards landscape start to evolve as as we get more into this world.
• 14:02
  Speaker 0: but again all of this is going to take a bit of time.
• 14:04
  Speaker 0: thank
• 14:04
  Speaker 2: you agree?
• 14:05
  Speaker 2: um what experiment or failure taught you the most about air technology transformation?
• 14:12
  Speaker 0: I would say that it's probably failure!
• 14:16
  Speaker 0: You know they say to kind of fail fast.
• 14:18
  Speaker 0: when first started with AI as a company.
• 14:22
  Speaker 0: We jumped in with a few POCs, you know this is going back couple of years at least now and we thought right what should we do?
• 14:30
  Speaker 0: And I was talking to two different AI consultancies and one were saying upscale everyone make your own AI literate and fluent.
• 14:36
  Speaker 0: set them free see what they can do.
• 14:38
  Speaker 0: there's another company that said let solve the specific problem using AI within the business to prove that it can be done.
• 14:44
  Speaker 0: And I thought, That was possibly the most sensible approach at the time.
• 14:48
  Speaker 0: so that's one we went with.
• 14:49
  Speaker 0: We developed a couple of PICs and proved what AI could do.
• 14:52
  Speaker 0: Then turned into how do we implement these in practice?
• 15:01
  Speaker 0: And we had to essentially stop the POCs at the end of their initial phase and go back to the drawing board.
• 15:06
  Speaker 0: What we realized as part learning is there's actually significant foundational work that needs to be done, to unlock real value from AI.
• 15:13
  Speaker 0: So parts of it are governance.
• 15:15
  Speaker 0: We've put in place good governance and risk management into our AI processes.
• 15:20
  Speaker 0: Parts of it is SOPs and controls for things like software development and technology management etc.
• 15:28
  Speaker 0: Also access data.
• 15:29
  Speaker 0: Every company has lots of data, but usually that data is locked away within a variety different systems.
• 15:36
  Speaker 0: It's very easy if you download data into an AI workspace to prove out and I use case improve it at work really hard to deploy enterprise grade solutions at pace in scale where I have access to source systems and can answer questions with specific privileges in real time.
• 15:53
  Speaker 0: That's some of the work that we've been doing this year to make sure that we can progress.
• 15:56
  Speaker 0: But in terms of my learning, it was failure from the first few realising the size of the task.
• 16:05
  Speaker 2: Thank you for sharing that experience as well as learning lessons.
• 16:08
  Speaker 2: That's very insightful, okay?
• 16:11
  Speaker 0: And
• 16:11
  Speaker 2: with that I think i'll pass it to Lacey for additional questions.
• 16:15
  Speaker 2: Yep
• 16:16
  Speaker 1: um let's see.
• 16:16
  Speaker 1: so as a EVP of data.
• 16:19
  Speaker 1: innovation in AI how do you personally decide which technologies to champion and which to wait on and which ones to buy out-of the box?
• 16:28
  Speaker 0: Well again good question especially in the AI landscape.
• 16:32
  Speaker 0: I think the landscape is changing so fast that it's sometimes better to wait and see, we talk about second mover advantage And i think thats definitely true In AI.
• 16:41
  Speaker 0: Sometimes you can see a gap.
• 16:43
  Speaker 0: We have seen gaps on AI technology.
• 16:45
  Speaker 0: We said alright well do this next sprint or project?
• 16:48
  Speaker 0: By which point?
• 16:49
  Speaker 0: That technology has been released or engineered into some of more generic solutions And if there are developments that are there, then it's kind of often no point in bespoking them yourselves.
• 17:00
  Speaker 0: So we've had to make decisions about whether to create a bespoke agent harnesses or where does wait for technology catch up?
• 17:06
  Speaker 0: Or weather two use the technologies built into generic LLM offerings.
• 17:12
  Speaker 0: generally my approach is that the key as if its a generic capability than waiters probably gonna come.
• 17:18
  Speaker 0: so if something all businesses and world are going to need at some point, then someone's gonna have the same idea and someone is going to go off and develop it.
• 17:26
  Speaker 0: And developing yourself would be a full zero.
• 17:29
  Speaker 0: but if its specific technology or data asset of which we had many in biometric than I think there value impressing ahead.
• 17:37
  Speaker 0: so when were prioritising spend on what we do?
• 17:40
  Speaker 0: We look for specific technologies that don't exist right now unlikely to exist in the public domain and we go ahead and develop those.
• 17:48
  Speaker 0: Very cool,
• 17:49
  Speaker 1: uh through its ups and downs who have always had the threat that jobs will go somewhere or go away with technology?
• 17:55
  Speaker 1: people working on the line In Michigan for instance becoming redundant to a robot, and then a robot in another country for example.
• 18:02
  Speaker 1: So let's address the elephant-in-the room there.
• 18:05
  Speaker 1: how do we make this AI moment?
• 18:07
  Speaker 1: A better one for ourselves?
• 18:09
  Speaker 1: is they're are positive way to interact.
• 18:11
  Speaker 1: embrace AI technology.
• 18:13
  Speaker 1: go along for their ride.
• 18:14
  Speaker 1: How can use it?
• 18:16
  Speaker 1: our personal professional development?
• 18:18
  Speaker 1: have you done that yet your own life?
• 18:21
  Speaker 0: yes There's a lot of yeses there, but I think in terms the jobs narrative which is The Elephant In A Room if you keep up with the current narrative.
• 18:31
  Speaker 0: The job discussion keeps changing.
• 18:33
  Speaker 0: it can't keep changing from doom and gloom to production uplift To um uh truth that people are never working at kind of their full productive capacity.
• 18:45
  Speaker 0: i mean If I look At my own to-do list I normally scratch twenty five to fifty percent Of It at best And That might be an overstatement.
• 18:52
  Speaker 0: As a result of using AI, I've become personally more productive.
• 18:55
  Speaker 0: And therefore doing more work and not do less time so does my job.
• 19:00
  Speaker 0: it didn't seem to be going away.
• 19:01
  Speaker 0: just seems to working harder than i was before potentially doing more the stuff that i did have time to do for.
• 19:08
  Speaker 0: The truth is no one really knows.
• 19:10
  Speaker 0: this narrative out there says some jobs will be potentially more susceptible then others and comes down how easy unbundles specific tasks from a role, um... From the human elements and the specific algorithmic elements that you could potentially pass over to some kind of AI.
• 19:32
  Speaker 0: And that's quite difficult!
• 19:33
  Speaker 0: That is a relatively new concept we're struggling with.
• 19:37
  Speaker 0: Struggling to unpick.
• 19:37
  Speaker 0: There are also the narrative that new jobs will be created.
• 19:42
  Speaker 0: We see it in our own business In terms of DevOps engineers The capacity for agent management training and deployment of solutions, which is ultimately just creating more work.
• 19:56
  Speaker 0: And I think that as we start to implement more AI it doesn't come for free.
• 20:02
  Speaker 0: It's not like switching on a light switch and suddenly you have productivity gains or cost savings.
• 20:08
  Speaker 0: i think there's going to be a huge lift for every organization To get the benefit from AI.
• 20:12
  Speaker 0: so certainly for the next few years at least um...I actually think There's gonna be more work rather than less As people try take the benefit from this technology.
• 20:22
  Speaker 0: In terms of my advice for what people do about that, I think My general advice would be This is not a bubble.
• 20:30
  Speaker 0: So my best advice would to get on the train And The best way To Do That Is Upskill Personally and Professionally.
• 20:38
  Speaker 0: There will never Be A Replacement For Relational Human Connection.
• 20:42
  Speaker 0: Indeed An AI Enabled Future The human connection Ironically becomes More Important.
• 20:49
  Speaker 0: But a hundred percent.
• 20:50
  Speaker 0: what I would suggest to people is use AI in their day-to-day world, where they can and whether you've got access that technology.
• 20:57
  Speaker 0: And start to learn because the learning curve does take some time and we have to go through several iterations before you start getting value from them.
• 21:04
  Speaker 0: as of me my personal development.
• 21:06
  Speaker 0: A hundred percent easy for myself.
• 21:08
  Speaker 0: I used it to be more organized.
• 21:10
  Speaker 0: Easy to plan better or easy to learn an acquire knowledge.
• 21:18
  Speaker 0: personality twin that I use to brainstorm with and i used to hold me into account.
• 21:23
  Speaker 0: And help me develop, deliver more from my role.
• 21:26
  Speaker 0: so you know personally...I've got a lot from it!
• 21:29
  Speaker 0: It's very exciting space for me but yeah..i think its available to everyone out there just if few words in the chat bot away.
• 21:36
  Speaker 1: That's right!
• 21:37
  Speaker 1: Its true....its like really getting into my life on daily basis You know?
• 21:42
  Speaker 1: It is SO
• 21:43
  Speaker 1: useful!!
• 21:44
  Speaker 1: We use them everything now.
• 21:46
  Speaker 1: Alright anyways, make it back
• 21:48
  Speaker 0: on top.
• 21:48
  Speaker 0: My kids actually my kids also at bedtime.
• 21:51
  Speaker 0: they know that you can talk into the phone because I've got a personal license of chat gpt and You can talk him to the phone and ask questions.
• 21:58
  Speaker 0: so when i'm doing bed time stories often my Six-year old wants to know what's the biggest chocolate bar ever made or was the tallest person that ever lived?
• 22:08
  Speaker 0: And we sit there We answer questions together all week we pull up pictures of sharks Mm-hmm find out all this kind of information.
• 22:16
  Speaker 0: So it's kind of finding its way into my professional life, but also in to my personal life and my children's lives as well.
• 22:21
  Speaker 1: Mm-Hmm a good strategy to move bedtime out a little further as well.
• 22:26
  Speaker 0: Yeah quite right An infinite number questions are there
• 22:30
  Speaker 1: exactly?
• 22:31
  Speaker 1: There's a great quote about how Culture each strategy for breakfast.
• 22:36
  Speaker 1: I love that quote And i feel like with the AI topic people.
• 22:42
  Speaker 1: the more people understand that it's accepted in our culture and we trust people to use it their work, I think they'll be more activated.
• 22:51
  Speaker 1: Just out of curiosity how do you feel?
• 22:54
  Speaker 1: We could better embed The Use Of AI In Our Culture At VeraMed so That People Feel Empowered To Use It And Make Their Jobs And Daily Lives More Enriched With AI.
• 23:09
  Speaker 0: Yeah, I think this is a really good question and it's something that has come up several times over the course of the last couple years at Veramed.
• 23:16
  Speaker 0: Now, Veramed is very high culture company.
• 23:20
  Speaker 0: we focus on people when you put people in the center with what they do.
• 23:24
  Speaker 0: so as a result of that where are companies?
• 23:27
  Speaker 0: Over indexes on human connection bring AI into their equation And there have been some fear.
• 23:35
  Speaker 0: resistance not just general technological fear that you might get in an explanation of crossing the chasm where you're talking about, um...the battle between early adopters and the early majority.
• 23:47
  Speaker 0: I think there's a resistance from a cultural perspective to it.
• 23:51
  Speaker 0: through the use of AI You can begin to become disingenuous Um..and produce content or interactions That are A lot less human than we used around here.
• 24:06
  Speaker 0: now.
• 24:07
  Speaker 0: I don't necessarily see it that way.
• 24:09
  Speaker 0: I, i don't see AI as a replacement for human connection and I've seen but As I said earlier in the discussion ,I think through the use of AI relational behaviors And Human Connection becomes even more important an Even More Powerful Than It Was Before.
• 24:27
  Speaker 0: Personally I See AI Is An Amplifier Of People's Skills & Abilities.
• 24:32
  Speaker 0: It Allows Them To Have Superpowers That They Didn'T have before.
• 24:37
  Speaker 0: But in no terms, and in no situations is that designed to act as a replacement for people.
• 24:44
  Speaker 0: So I think what we encourage people to do—and what i would encourage everyone listening to this podcast to do —is make sure when they're using it AI especially as it relates to connection between themselves and others To make sure that stays as human as possible.
• 25:03
  Speaker 0: That doesn't mean you can't use AI help you draft an email I often do if i can't think about how to phrase something, but it means make sure that you read That email.
• 25:12
  Speaker 0: Make sure that it sounds like you made sure that You put yourselves in the shoes of the receiver.
• 25:19
  Speaker 0: For example even removing m dashes from an e-mail Crazy so there is kind of a eyes.
• 25:26
  Speaker 0: The hallmarks of AI aren't left within your message And then I think they're back and add a lot.
• 25:32
  Speaker 0: But that being said I think ai's starting to become they're more generally accepted, provided as I said is that people don't use it as a kind of complete replacement for their own brain and the use of that looks good.
• 25:47
  Speaker 0: But what i'm seeing in baromet is that there are people who are using it responsibly and they're using it culturally adept way.
• 25:55
  Speaker 0: where that's not been the case we're sensitively handling Mm-hmm.
• 26:01
  Speaker 1: Well as podcast hosts, it's good to have a human interaction with you about this topic.
• 26:05
  Speaker 1: so You know I'm thinking that the AI isn't going to replace the podcast team anytime soon.
• 26:11
  Speaker 1: So that's That's a good thing and I feel like... You
• 26:14
  Speaker 0: never know though, Lacey Never know because i listen to many podcasts And there is quite a lot developed and spoken by AI.
• 26:21
  Speaker 1: Oh god!
• 26:23
  Speaker 0: Some of them are quite good.
• 26:25
  Speaker 0: The key though Is context If the content created and curated in the right way by humans so that equality is kind of much higher than generic LLM output.
• 26:38
  Speaker 0: It can be quite strong, but as you say I don't think there's any replacement for human connection.
• 26:43
  Speaker 0: So i really hope we don't get replaced anytime soon.
• 26:48
  Speaker 1: Agreed okay?
• 26:49
  Speaker 1: The final question then one everyone has probably appeared for Team Cat or Team Dog.
• 26:56
  Speaker 0: Well I would sit on the fence, but i am officially team cat.
• 27:01
  Speaker 0: The reason for this is that one of my children is allergic to dogs.
• 27:05
  Speaker 0: so as a family we're officially Team Cat and we are the proud owner of a cat who was rescued by a very med team member And then adopted about year ago.
• 27:15
  Speaker 0: So
• 27:17
  Speaker 1: congratulations on the new cat!
• 27:18
  Speaker 1: That's awesome Lucky cat.
• 27:23
  Speaker 1: Okay, Rashali I'm gonna pass it back to you.
• 27:25
  Speaker 2: Well just before we wrap up this session more like a personal question if You get an opportunity To have dinner with three people of your choice either dead or Aligned whom would you choose and why?
• 27:46
  Speaker 0: Now this question.
• 27:47
  Speaker 0: i knew This question was coming but so long and hard about It And they definitely wouldn't be AI avatars that would come to my dinner party.
• 27:57
  Speaker 0: In all seriousness, I've got a group of really close friends but they're distributed all over the world.
• 28:04
  Speaker 0: as far afield is East and West Australia New Zealand other places etc.
• 28:09
  Speaker 0: so it will be them.
• 28:11
  Speaker 0: It's been about ten years since we all sat down for dinner together So you might be a bit sad But i'll probably pick then.
• 28:18
  Speaker 2: That sounds good Okay And on that note
• 28:22
  Speaker 1: Now
• 28:22
  Speaker 2: we can wrap up things for today.
• 28:25
  Speaker 2: Matt, thank you so much for an open and thought-provoking conversation.
• 28:30
  Speaker 2: It's really been very insightful to explore not just the potential of AI but how we can approach it in a responsible and human looped way.
• 28:42
  Speaker 2: Thank You So Much For Joining us on this podcast journey.
• 28:46
  Speaker 2: We hope Todays Conversation has sparked few ideas,
• 28:50
  Speaker 1: as well
• 28:51
  Speaker 2: as curiosity to explore further in our own work area.
• 28:56
  Speaker 1: And
• 28:57
  Speaker 2: don't forget to tune-in for next episode where we'll continue the conversations and
• 29:02
  Speaker 1: dive
• 29:02
  Speaker 2: into more interesting topics so
• 29:08
  Speaker 0: until.