Up next in 10
The Cloud Show with Magnus Mårtensson ft. Olena Borzenko - Ep: 64
0 views
Aug 6, 2025
Join this live session with Magnus Mårtensson ft. Olena Borzenko for the next episode of The Cloud Show with Magnus Mårtensson on March 05 at 01:05 PM (EST). The show is about cloud leadership and all the important questions relating to cloud projects. Certainly, many matters when a company is going to and wants to be successful in the cloud, are about technology. However, there are many additional matters, adjacent to technology, that we also need to tend to regarding business strategy, human resources, organizational change, planning for a technical cloud approach, and many more questions. These conversations are critical for a healthy cloud and for a swift and accurate cloud approach. GUEST SPEAKER Senior backend developer with a Bachelor's degree focused in Engineering Physics/Applied Physics from Lviv Polytechnic National University. 📺 CSharp TV - Dev Streaming Destination http://csharp.tv 🌎 C# Corner - Community of Software and Data Developers https://www.c-sharpcorner.com #CSharpTV #csharpcorner #TheCloudShow #CSharp #Interview
View Video Transcript
0:04
hello everyone welcome back to another
0:06
hello everyone welcome back to another
0:06
hello everyone welcome back to another episode of the cloud show and again we
0:08
episode of the cloud show and again we
0:08
episode of the cloud show and again we are on the road traveling the cloud show
0:10
are on the road traveling the cloud show
0:10
are on the road traveling the cloud show to a different place a different
0:12
to a different place a different
0:12
to a different place a different continent again I am down in Tunisia and
0:16
continent again I am down in Tunisia and
0:16
continent again I am down in Tunisia and I had the opportunity to catch up with
0:18
I had the opportunity to catch up with
0:18
I had the opportunity to catch up with an old friend and make sure to get a
0:20
an old friend and make sure to get a
0:21
an old friend and make sure to get a first second appearance on my show no
0:24
first second appearance on my show no
0:24
first second appearance on my show no one else has been on the show twice
0:25
one else has been on the show twice
0:26
one else has been on the show twice except this time I'm going to sit down
0:28
except this time I'm going to sit down
0:28
except this time I'm going to sit down again with Elena and we are going talk
0:30
again with Elena and we are going talk
0:30
again with Elena and we are going talk about Ai and ethics on the cloud
0:33
about Ai and ethics on the cloud
0:33
about Ai and ethics on the cloud [Music]
0:40
[Music]
0:40
[Music] show hello Elena how are you hi I'm
0:44
show hello Elena how are you hi I'm
0:44
show hello Elena how are you hi I'm really good you made it sound like a
0:46
really good you made it sound like a
0:46
really good you made it sound like a really big deal being the second time in
0:47
really big deal being the second time in
0:47
really big deal being the second time in the show yeah but big honor yeah for
0:50
the show yeah but big honor yeah for
0:50
the show yeah but big honor yeah for having me again it's got to it's got to
0:51
having me again it's got to it's got to
0:51
having me again it's got to it's got to happen sometime and it couldn't have
0:53
happen sometime and it couldn't have
0:53
happen sometime and it couldn't have been a better person so you're the first
0:55
been a better person so you're the first
0:55
been a better person so you're the first second appearance on the cloud show
0:57
second appearance on the cloud show
0:57
second appearance on the cloud show thank you and here at iian de days it's
1:01
thank you and here at iian de days it's
1:01
thank you and here at iian de days it's pretty good conference right yeah I
1:03
pretty good conference right yeah I
1:03
pretty good conference right yeah I really liked it enjoyed it so far really
1:04
really liked it enjoyed it so far really
1:04
really liked it enjoyed it so far really cool questions really cool people so
1:07
cool questions really cool people so
1:07
cool questions really cool people so really nice place I want to go here come
1:08
really nice place I want to go here come
1:08
really nice place I want to go here come here again next I would I would go I
1:10
here again next I would I would go I
1:10
here again next I would I would go I would come a second time as well yeah
1:12
would come a second time as well yeah
1:12
would come a second time as well yeah yeah definitely no no they're doing a
1:13
yeah definitely no no they're doing a
1:13
yeah definitely no no they're doing a great job and uh apparently it's it's
1:16
great job and uh apparently it's it's
1:16
great job and uh apparently it's it's kind of unusual to have uh Tech
1:18
kind of unusual to have uh Tech
1:18
kind of unusual to have uh Tech conferences in Tunisia so we're kind of
1:20
conferences in Tunisia so we're kind of
1:20
conferences in Tunisia so we're kind of happy to be here for the for the start
1:22
happy to be here for the for the start
1:22
happy to be here for the for the start of this yeah start yeah start up
1:25
of this yeah start yeah start up
1:25
of this yeah start yeah start up something good all right so you had a
1:27
something good all right so you had a
1:27
something good all right so you had a session here of course that's why you're
1:29
session here of course that's why you're
1:29
session here of course that's why you're here ultimately and you were talking
1:31
here ultimately and you were talking
1:32
here ultimately and you were talking about Ai and ethics yeah it's a very
1:35
about Ai and ethics yeah it's a very
1:35
about Ai and ethics yeah it's a very current topic yeah very sensitive topic
1:38
current topic yeah very sensitive topic
1:38
current topic yeah very sensitive topic very yeah yeah so I I guess in this
1:41
very yeah yeah so I I guess in this
1:42
very yeah yeah so I I guess in this space it's super important that we are
1:44
space it's super important that we are
1:44
space it's super important that we are able to do the right thing the
1:46
able to do the right thing the
1:46
able to do the right thing the responsible thing the correct thing with
1:49
responsible thing the correct thing with
1:49
responsible thing the correct thing with AI because AI can definitely be abused
1:53
AI because AI can definitely be abused
1:53
AI because AI can definitely be abused and it can have some um unwanted results
1:57
and it can have some um unwanted results
1:57
and it can have some um unwanted results unwanted outcomes so shall we start
1:59
unwanted outcomes so shall we start
1:59
unwanted outcomes so shall we start there if you will uh what are some
2:02
there if you will uh what are some
2:02
there if you will uh what are some problems that AI can have uh today um
2:06
problems that AI can have uh today um
2:06
problems that AI can have uh today um you know there's quite some errors like
2:08
you know there's quite some errors like
2:08
you know there's quite some errors like multiple errors um that are impacting
2:11
multiple errors um that are impacting
2:11
multiple errors um that are impacting the eye the results the out outcome and
2:13
the eye the results the out outcome and
2:13
the eye the results the out outcome and how we use it but I think in my opinion
2:15
how we use it but I think in my opinion
2:15
how we use it but I think in my opinion there is a three key points that I can
2:17
there is a three key points that I can
2:17
there is a three key points that I can mention so first of all of all it's a
2:20
mention so first of all of all it's a
2:20
mention so first of all of all it's a data that we using that been used for
2:22
data that we using that been used for
2:22
data that we using that been used for training the models the second one is
2:25
training the models the second one is
2:25
training the models the second one is transparency and accuracy it's basically
2:28
transparency and accuracy it's basically
2:28
transparency and accuracy it's basically uh the way how it decision is being made
2:32
uh the way how it decision is being made
2:32
uh the way how it decision is being made when we use the for certain tasks for
2:34
when we use the for certain tasks for
2:34
when we use the for certain tasks for example uh to summarize the certain
2:37
example uh to summarize the certain
2:37
example uh to summarize the certain documents for example or using it for
2:39
documents for example or using it for
2:39
documents for example or using it for hiring tools how those decision is being
2:41
hiring tools how those decision is being
2:41
hiring tools how those decision is being made there's a data set that's being
2:44
made there's a data set that's being
2:44
made there's a data set that's being used but also there's algorithms so
2:46
used but also there's algorithms so
2:46
used but also there's algorithms so that's the second aspect and the third
2:48
that's the second aspect and the third
2:48
that's the second aspect and the third aspect is how we actually integrating
2:50
aspect is how we actually integrating
2:50
aspect is how we actually integrating those tools in the companies one thing
2:52
those tools in the companies one thing
2:52
those tools in the companies one thing is having a great data and great
2:54
is having a great data and great
2:54
is having a great data and great algorithm but if you misuse it that's on
2:57
algorithm but if you misuse it that's on
2:57
algorithm but if you misuse it that's on you that's your responsibility right
2:59
you that's your responsibility right
2:59
you that's your responsibility right right so so if I get this right it's
3:01
right so so if I get this right it's
3:01
right so so if I get this right it's it's having data that has problems bias
3:04
it's having data that has problems bias
3:04
it's having data that has problems bias maybe it's it's how we then process that
3:07
maybe it's it's how we then process that
3:07
maybe it's it's how we then process that data the algorithm that actually does
3:09
data the algorithm that actually does
3:09
data the algorithm that actually does the processing of the data and then it
3:12
the processing of the data and then it
3:12
the processing of the data and then it is how we users actually use the data or
3:16
is how we users actually use the data or
3:16
is how we users actually use the data or the the the capability ultimately how we
3:19
the the the capability ultimately how we
3:19
the the the capability ultimately how we how we use it and how we integrate it
3:20
how we use it and how we integrate it
3:20
how we use it and how we integrate it right yeah I've been given a lot of
3:22
right yeah I've been given a lot of
3:23
right yeah I've been given a lot of examples in this space and one one that
3:25
examples in this space and one one that
3:25
examples in this space and one one that comes to mind is for example if you have
3:27
comes to mind is for example if you have
3:27
comes to mind is for example if you have a chatbot which everybody is supposed to
3:29
a chatbot which everybody is supposed to
3:29
a chatbot which everybody is supposed to have now I guess uh and and uh the
3:32
have now I guess uh and and uh the
3:32
have now I guess uh and and uh the chatbot typically for a company should
3:34
chatbot typically for a company should
3:34
chatbot typically for a company should be talking about the company's product
3:35
be talking about the company's product
3:35
be talking about the company's product that that would be nice right but but
3:38
that that would be nice right but but
3:38
that that would be nice right but but then we also know that if we ask the
3:40
then we also know that if we ask the
3:40
then we also know that if we ask the chatbot different other things then it
3:42
chatbot different other things then it
3:42
chatbot different other things then it it might still respond to those like for
3:44
it might still respond to those like for
3:44
it might still respond to those like for example oh I'm a little bit hungry and
3:46
example oh I'm a little bit hungry and
3:46
example oh I'm a little bit hungry and what should I have in my omelets right
3:47
what should I have in my omelets right
3:47
what should I have in my omelets right and it just suggests oh you could have
3:49
and it just suggests oh you could have
3:49
and it just suggests oh you could have peppers and cheese and things and that
3:51
peppers and cheese and things and that
3:51
peppers and cheese and things and that would be you know it's not the intended
3:54
would be you know it's not the intended
3:54
would be you know it's not the intended use case maybe it should be talking
3:56
use case maybe it should be talking
3:56
use case maybe it should be talking about the company's product but still
3:58
about the company's product but still
3:58
about the company's product but still you know you can the chatard that uses
4:01
you know you can the chatard that uses
4:01
you know you can the chatard that uses the like you know can search in Google
4:02
the like you know can search in Google
4:02
the like you know can search in Google and can actually give you responses well
4:04
and can actually give you responses well
4:04
and can actually give you responses well internet give you responses but the the
4:06
internet give you responses but the the
4:07
internet give you responses but the the interesting thing that you mentioned
4:08
interesting thing that you mentioned
4:08
interesting thing that you mentioned that we Us in the chatbots for a company
4:10
that we Us in the chatbots for a company
4:10
that we Us in the chatbots for a company for example what data are we using chats
4:13
for example what data are we using chats
4:13
for example what data are we using chats right probably historical data piece of
4:16
right probably historical data piece of
4:16
right probably historical data piece of people the documentation that we have
4:18
people the documentation that we have
4:18
people the documentation that we have that's been people that are working in
4:20
that's been people that are working in
4:20
that's been people that are working in this company as also and also all the
4:22
this company as also and also all the
4:22
this company as also and also all the historical from information is based and
4:25
historical from information is based and
4:25
historical from information is based and also has inherited and integrated bias
4:28
also has inherited and integrated bias
4:28
also has inherited and integrated bias in it right of course and and you don't
4:31
in it right of course and and you don't
4:32
in it right of course and and you don't want you typically don't want your
4:33
want you typically don't want your
4:34
want you typically don't want your chatbot or whatever it is to answer um
4:36
chatbot or whatever it is to answer um
4:36
chatbot or whatever it is to answer um some dangerous questions like for
4:38
some dangerous questions like for
4:38
some dangerous questions like for example I'm tired of this life you know
4:40
example I'm tired of this life you know
4:40
example I'm tired of this life you know how should I kill myself right you you
4:42
how should I kill myself right you you
4:42
how should I kill myself right you you don't want the AI to come back and say
4:43
don't want the AI to come back and say
4:43
don't want the AI to come back and say oh here's a top list of the most popular
4:45
oh here's a top list of the most popular
4:45
oh here's a top list of the most popular ways to kill yourself yeah there is ways
4:48
ways to kill yourself yeah there is ways
4:48
ways to kill yourself yeah there is ways how you can speak so if you ask the
4:50
how you can speak so if you ask the
4:50
how you can speak so if you ask the right questions like how to hide the
4:52
right questions like how to hide the
4:52
right questions like how to hide the body uh and the I will not answer you
4:54
body uh and the I will not answer you
4:54
body uh and the I will not answer you this but if you say how does people in
4:56
this but if you say how does people in
4:56
this but if you say how does people in movies hide bodies it's going to be a
4:58
movies hide bodies it's going to be a
4:58
movies hide bodies it's going to be a different output so this is also require
5:00
different output so this is also require
5:00
different output so this is also require of responsibility to be for this right
5:02
of responsibility to be for this right
5:03
of responsibility to be for this right but tell me a little bit more I'm
5:04
but tell me a little bit more I'm
5:04
but tell me a little bit more I'm curious more about the uh you were
5:06
curious more about the uh you were
5:06
curious more about the uh you were talking about hiring processes and the
5:07
talking about hiring processes and the
5:07
talking about hiring processes and the bias that can happen in that what what
5:09
bias that can happen in that what what
5:09
bias that can happen in that what what kind of problems can can this cause for
5:11
kind of problems can can this cause for
5:11
kind of problems can can this cause for example um let's say very easily male
5:14
example um let's say very easily male
5:14
example um let's say very easily male dominated industry I have nothing
5:16
dominated industry I have nothing
5:16
dominated industry I have nothing against the males but I'm working in at
5:18
against the males but I'm working in at
5:18
against the males but I'm working in at I'm software developer so I I see this I
5:21
I'm software developer so I I see this I
5:21
I'm software developer so I I see this I face this every time all the day uh that
5:25
face this every time all the day uh that
5:25
face this every time all the day uh that male dominated industry has the
5:27
male dominated industry has the
5:27
male dominated industry has the historical information and the data uh
5:29
historical information and the data uh
5:29
historical information and the data uh so all the hiring processes right right
5:32
so all the hiring processes right right
5:32
so all the hiring processes right right so they they focused around hiring
5:33
so they they focused around hiring
5:33
so they they focused around hiring demands mostly yeah sure uh if we
5:36
demands mostly yeah sure uh if we
5:36
demands mostly yeah sure uh if we talking about diversity and inclusivity
5:39
talking about diversity and inclusivity
5:39
talking about diversity and inclusivity and all the stuff so companies have this
5:41
and all the stuff so companies have this
5:41
and all the stuff so companies have this also on the agenda we want to make
5:43
also on the agenda we want to make
5:43
also on the agenda we want to make companies more diverse more inclusive
5:45
companies more diverse more inclusive
5:45
companies more diverse more inclusive you want to hire more women but then
5:47
you want to hire more women but then
5:47
you want to hire more women but then they integrate in in their hiring
5:49
they integrate in in their hiring
5:49
they integrate in in their hiring processes and they don't adjust the data
5:51
processes and they don't adjust the data
5:51
processes and they don't adjust the data that they provide so they provide all
5:53
that they provide so they provide all
5:53
that they provide so they provide all the successful uh hiring cases from the
5:56
the successful uh hiring cases from the
5:56
the successful uh hiring cases from the historical information and it's because
5:58
historical information and it's because
5:58
historical information and it's because it's the industry most of the cases will
6:01
it's the industry most of the cases will
6:01
it's the industry most of the cases will be male uh male cases yeah so hiring
6:05
be male uh male cases yeah so hiring
6:05
be male uh male cases yeah so hiring jobs and that makes sense yeah that's
6:07
jobs and that makes sense yeah that's
6:07
jobs and that makes sense yeah that's really makes a big problem because then
6:09
really makes a big problem because then
6:10
really makes a big problem because then you will be just like your c will be
6:12
you will be just like your c will be
6:12
you will be just like your c will be just um skipped because uh certain
6:16
just um skipped because uh certain
6:16
just um skipped because uh certain aspects and it can be gender if you if
6:18
aspects and it can be gender if you if
6:18
aspects and it can be gender if you if you have many men have been successfully
6:21
you have many men have been successfully
6:21
you have many men have been successfully hired before and that's the data we have
6:24
hired before and that's the data we have
6:24
hired before and that's the data we have and then if you put a a a woman's CV in
6:27
and then if you put a a a woman's CV in
6:27
and then if you put a a a woman's CV in the pile it's going to be like
6:30
the pile it's going to be like
6:30
the pile it's going to be like understand those CVS they have
6:32
understand those CVS they have
6:32
understand those CVS they have information like gender nationality and
6:36
information like gender nationality and
6:36
information like gender nationality and this might not be craved or or like you
6:38
this might not be craved or or like you
6:38
this might not be craved or or like you know uh integrated in the algorithm
6:40
know uh integrated in the algorithm
6:40
know uh integrated in the algorithm itself but because AI has access to
6:42
itself but because AI has access to
6:42
itself but because AI has access to certain data right so it's not the
6:44
certain data right so it's not the
6:44
certain data right so it's not the company's policy it's not what they want
6:47
company's policy it's not what they want
6:47
company's policy it's not what they want and they have no no you know if you talk
6:49
and they have no no you know if you talk
6:50
and they have no no you know if you talk to people they they would like to hire
6:51
to people they they would like to hire
6:51
to people they they would like to hire more women but then they maybe use a
6:54
more women but then they maybe use a
6:54
more women but then they maybe use a tool which has a bias right and maybe
6:57
tool which has a bias right and maybe
6:57
tool which has a bias right and maybe accidentally but still the outcome is
6:59
accidentally but still the outcome is
6:59
accidentally but still the outcome is not what they what they wanted right but
7:01
not what they what they wanted right but
7:01
not what they what they wanted right but it's also we cannot say that it's the
7:03
it's also we cannot say that it's the
7:03
it's also we cannot say that it's the company has no responsibility over it oh
7:06
company has no responsibility over it oh
7:06
company has no responsibility over it oh because they do there are ways even if
7:08
because they do there are ways even if
7:08
because they do there are ways even if you have the historical information
7:10
you have the historical information
7:10
you have the historical information there bias you can still avoid this bias
7:13
there bias you can still avoid this bias
7:13
there bias you can still avoid this bias in your responses that's the I think the
7:15
in your responses that's the I think the
7:15
in your responses that's the I think the important part they maybe didn't want to
7:18
important part they maybe didn't want to
7:18
important part they maybe didn't want to have this outcome um they were not
7:19
have this outcome um they were not
7:19
have this outcome um they were not looking for such an outcome and they
7:22
looking for such an outcome and they
7:22
looking for such an outcome and they they they made the wrong choice along
7:24
they they made the wrong choice along
7:24
they they made the wrong choice along the way right they they they couldn't
7:26
the way right they they they couldn't
7:26
the way right they they they couldn't use the AI to help them instead the AI
7:30
use the AI to help them instead the AI
7:30
use the AI to help them instead the AI did the well the AI is always going to
7:32
did the well the AI is always going to
7:32
did the well the AI is always going to do its best right based on the
7:33
do its best right based on the
7:33
do its best right based on the information that it has it will do its
7:35
information that it has it will do its
7:35
information that it has it will do its very best job but it if it has wrong
7:37
very best job but it if it has wrong
7:37
very best job but it if it has wrong information then the answer will not be
7:39
information then the answer will not be
7:39
information then the answer will not be the desired outcome yeah right I think
7:41
the desired outcome yeah right I think
7:41
the desired outcome yeah right I think it's very important for us as humans to
7:44
it's very important for us as humans to
7:44
it's very important for us as humans to work alongside with so and guide the
7:46
work alongside with so and guide the
7:46
work alongside with so and guide the process of uh decision making so we can
7:49
process of uh decision making so we can
7:49
process of uh decision making so we can say it's our responsibility to say uh
7:52
say it's our responsibility to say uh
7:52
say it's our responsibility to say uh even when we creating the prompt try to
7:54
even when we creating the prompt try to
7:54
even when we creating the prompt try to have a better outlook on the picture to
7:56
have a better outlook on the picture to
7:56
have a better outlook on the picture to certain factors don't take into account
7:59
certain factors don't take into account
7:59
certain factors don't take into account gender race nationality you can create a
8:01
gender race nationality you can create a
8:01
gender race nationality you can create a safe guard trails with safe rules to
8:04
safe guard trails with safe rules to
8:04
safe guard trails with safe rules to make sure there's less bonus okay so so
8:07
make sure there's less bonus okay so so
8:07
make sure there's less bonus okay so so let's let's let's focus in on that and
8:09
let's let's let's focus in on that and
8:09
let's let's let's focus in on that and see so what what are some things that we
8:11
see so what what are some things that we
8:11
see so what what are some things that we can do to as AI or we want to use AI
8:16
can do to as AI or we want to use AI
8:16
can do to as AI or we want to use AI right what are some things that we can
8:17
right what are some things that we can
8:17
right what are some things that we can do to uh make sure that we get an
8:20
do to uh make sure that we get an
8:20
do to uh make sure that we get an unbiased result or we get uh you know to
8:23
unbiased result or we get uh you know to
8:23
unbiased result or we get uh you know to a a fair selection and and or you know
8:27
a a fair selection and and or you know
8:27
a a fair selection and and or you know what do we do let's have a look at a
8:29
what do we do let's have a look at a
8:29
what do we do let's have a look at a case so of course there's algorithms
8:31
case so of course there's algorithms
8:32
case so of course there's algorithms between in data we don't have control
8:33
between in data we don't have control
8:34
between in data we don't have control over this as a final user I still can do
8:37
over this as a final user I still can do
8:37
over this as a final user I still can do something so uh if I'm having a company
8:39
something so uh if I'm having a company
8:39
something so uh if I'm having a company and I want to make sure that in of my
8:41
and I want to make sure that in of my
8:41
and I want to make sure that in of my company using responsibly I will make
8:43
company using responsibly I will make
8:43
company using responsibly I will make sure that they getting uh correct enough
8:46
sure that they getting uh correct enough
8:46
sure that they getting uh correct enough education about this okay so where the
8:49
education about this okay so where the
8:49
education about this okay so where the data is coming from uh if it's coming
8:51
data is coming from uh if it's coming
8:51
data is coming from uh if it's coming from underrepresented groups for example
8:53
from underrepresented groups for example
8:53
from underrepresented groups for example I want to make sure that when they
8:55
I want to make sure that when they
8:55
I want to make sure that when they making the request of this data they
8:57
making the request of this data they
8:57
making the request of this data they also set the certain rules and guard
8:59
also set the certain rules and guard
8:59
also set the certain rules and guard trails to avoid the bias in it so proper
9:02
trails to avoid the bias in it so proper
9:02
trails to avoid the bias in it so proper education yeah making sure raising the
9:05
education yeah making sure raising the
9:05
education yeah making sure raising the awareness about um potential risks and
9:08
awareness about um potential risks and
9:08
awareness about um potential risks and biases in the data and outcomes and also
9:12
biases in the data and outcomes and also
9:12
biases in the data and outcomes and also um well writing proper proms probably
9:15
um well writing proper proms probably
9:15
um well writing proper proms probably having some AI guidelines will help a
9:17
having some AI guidelines will help a
9:17
having some AI guidelines will help a lot in your company right right so
9:20
lot in your company right right so
9:20
lot in your company right right so making even some tools by developers who
9:23
making even some tools by developers who
9:23
making even some tools by developers who understand how AI works for the use of
9:25
understand how AI works for the use of
9:25
understand how AI works for the use of other people in the company can help a
9:27
other people in the company can help a
9:27
other people in the company can help a lot so it's we saw a lot exposion of
9:30
lot so it's we saw a lot exposion of
9:30
lot so it's we saw a lot exposion of different chat so you can even go to CH
9:32
different chat so you can even go to CH
9:32
different chat so you can even go to CH you can create your own chat right
9:35
you can create your own chat right
9:35
you can create your own chat right nothing stopping you from creating that
9:36
nothing stopping you from creating that
9:36
nothing stopping you from creating that one is the correct system message that
9:39
one is the correct system message that
9:39
one is the correct system message that will make sure that there's less biased
9:41
will make sure that there's less biased
9:41
will make sure that there's less biased results for example so you put all those
9:43
results for example so you put all those
9:43
results for example so you put all those guard trails in there so you you are
9:45
guard trails in there so you you are
9:45
guard trails in there so you you are careful to uh to instruct the AI to uh
9:50
careful to uh to instruct the AI to uh
9:50
careful to uh to instruct the AI to uh to adjust
9:51
to adjust
9:51
to adjust for a a a data set that isn't that isn't
9:55
for a a a data set that isn't that isn't
9:55
for a a a data set that isn't that isn't fair yeah we cannot avoid this because
9:58
fair yeah we cannot avoid this because
9:58
fair yeah we cannot avoid this because we cannot avoid any like predisposition
10:00
we cannot avoid any like predisposition
10:00
we cannot avoid any like predisposition or some because the people creating Ai
10:04
or some because the people creating Ai
10:04
or some because the people creating Ai and AI is training all the data provided
10:07
and AI is training all the data provided
10:07
and AI is training all the data provided by us that's right data so it will exist
10:10
by us that's right data so it will exist
10:10
by us that's right data so it will exist always but we have to make sure that
10:12
always but we have to make sure that
10:12
always but we have to make sure that when AI making decision it's not because
10:15
when AI making decision it's not because
10:15
when AI making decision it's not because of certain factors so do you think that
10:17
of certain factors so do you think that
10:17
of certain factors so do you think that we now using AI has have put a more
10:23
we now using AI has have put a more
10:23
we now using AI has have put a more focus on these questions than we have
10:26
focus on these questions than we have
10:26
focus on these questions than we have had historically because the the the bu
10:29
had historically because the the the bu
10:29
had historically because the the the bu in the data has been around since the
10:31
in the data has been around since the
10:31
in the data has been around since the the age of the internet right it's it's
10:33
the age of the internet right it's it's
10:33
the age of the internet right it's it's been dominated by people that may maybe
10:36
been dominated by people that may maybe
10:36
been dominated by people that may maybe look just like me like like white male
10:39
look just like me like like white male
10:39
look just like me like like white male middle-aged person right um has a lot of
10:43
middle-aged person right um has a lot of
10:43
middle-aged person right um has a lot of uh emphasis in the data to say right are
10:46
uh emphasis in the data to say right are
10:46
uh emphasis in the data to say right are we are we getting can AI maybe help us
10:49
we are we getting can AI maybe help us
10:49
we are we getting can AI maybe help us make this better in the future um I
10:52
make this better in the future um I
10:52
make this better in the future um I think we can work alongside so first we
10:55
think we can work alongside so first we
10:55
think we can work alongside so first we have to make sure that humans actually
10:57
have to make sure that humans actually
10:57
have to make sure that humans actually assist in eii and not like fighting
10:59
assist in eii and not like fighting
11:00
assist in eii and not like fighting against it when you fight it you fear it
11:02
against it when you fight it you fear it
11:02
against it when you fight it you fear it and you don't understand it if you
11:04
and you don't understand it if you
11:04
and you don't understand it if you assist it you understand and you guide
11:06
assist it you understand and you guide
11:06
assist it you understand and you guide and you show the direction that's the
11:08
and you show the direction that's the
11:08
and you show the direction that's the first thing and the second thing of
11:10
first thing and the second thing of
11:10
first thing and the second thing of course the bias data exist for a long
11:13
course the bias data exist for a long
11:13
course the bias data exist for a long time and that's very good examples is of
11:16
time and that's very good examples is of
11:16
time and that's very good examples is of healthare so we know a lot of examples
11:19
healthare so we know a lot of examples
11:19
healthare so we know a lot of examples of mistakes being made because of the
11:21
of mistakes being made because of the
11:21
of mistakes being made because of the data sets right even before CH and all
11:24
data sets right even before CH and all
11:24
data sets right even before CH and all this but I think because of the exposure
11:27
this but I think because of the exposure
11:28
this but I think because of the exposure of EI we have
11:30
of EI we have
11:30
of EI we have uh the problem became so big and so fast
11:32
uh the problem became so big and so fast
11:32
uh the problem became so big and so fast that we have to be much more uh like
11:35
that we have to be much more uh like
11:35
that we have to be much more uh like careful about it and put much more
11:37
careful about it and put much more
11:37
careful about it and put much more intention than before and in this space
11:40
intention than before and in this space
11:40
intention than before and in this space then that's is I think I read something
11:41
then that's is I think I read something
11:42
then that's is I think I read something about that that that we're working to
11:43
about that that that we're working to
11:43
about that that that we're working to create standards and policies for this
11:46
create standards and policies for this
11:46
create standards and policies for this kind of work right now two years after
11:49
kind of work right now two years after
11:49
kind of work right now two years after the app was launched got like in two
11:51
the app was launched got like in two
11:51
the app was launched got like in two months 100 million should have done that
11:53
months 100 million should have done that
11:53
months 100 million should have done that the other way around but at least we are
11:55
the other way around but at least we are
11:55
the other way around but at least we are working on it now to say that there are
11:58
working on it now to say that there are
11:58
working on it now to say that there are there are there are things we have have
11:59
there are there are things we have have
11:59
there are there are things we have have to consider when we're using AI there
12:01
to consider when we're using AI there
12:01
to consider when we're using AI there are some policies we have to follow yeah
12:03
are some policies we have to follow yeah
12:03
are some policies we have to follow yeah right makes sense kind of like first we
12:05
right makes sense kind of like first we
12:05
right makes sense kind of like first we do then we see the consequences and then
12:07
do then we see the consequences and then
12:07
do then we see the consequences and then we try to find a solution right so so
12:10
we try to find a solution right so so
12:10
we try to find a solution right so so the the the summary of the of the advice
12:12
the the the summary of the of the advice
12:13
the the the summary of the of the advice that we need as as a company where we a
12:15
that we need as as a company where we a
12:15
that we need as as a company where we a company we're going to use start using
12:16
company we're going to use start using
12:16
company we're going to use start using AI in summary we have to look at our
12:18
AI in summary we have to look at our
12:18
AI in summary we have to look at our data right to figure out if is this say
12:21
data right to figure out if is this say
12:21
data right to figure out if is this say what else data guines have guidelines
12:24
what else data guines have guidelines
12:24
what else data guines have guidelines for it yeah and you you talked about
12:26
for it yeah and you you talked about
12:26
for it yeah and you you talked about training uh yeah you can you can well
12:29
training uh yeah you can you can well
12:29
training uh yeah you can you can well the first is data ni guidelines and then
12:31
the first is data ni guidelines and then
12:31
the first is data ni guidelines and then you can have actually people creating
12:33
you can have actually people creating
12:33
you can have actually people creating tools that are being safe to use for uh
12:36
tools that are being safe to use for uh
12:36
tools that are being safe to use for uh people that are not that knowledgeable
12:37
people that are not that knowledgeable
12:37
people that are not that knowledgeable in the I for example okay okay right and
12:40
in the I for example okay okay right and
12:40
in the I for example okay okay right and also educating people of what comp is
12:43
also educating people of what comp is
12:43
also educating people of what comp is how to use risk have right right and
12:47
how to use risk have right right and
12:47
how to use risk have right right and then there are there are training
12:48
then there are there are training
12:48
then there are there are training materials out there for for for for
12:51
materials out there for for for for
12:51
materials out there for for for for learning for an organization to learn
12:53
learning for an organization to learn
12:53
learning for an organization to learn how to use AI in a responsible and a and
12:55
how to use AI in a responsible and a and
12:55
how to use AI in a responsible and a and a more safe way I I think you said
12:57
a more safe way I I think you said
12:57
a more safe way I I think you said trainings trainings would be defin think
13:00
trainings trainings would be defin think
13:00
trainings trainings would be defin think create um nonbiased prompt that's would
13:02
create um nonbiased prompt that's would
13:02
create um nonbiased prompt that's would be a great yeah good training quick one
13:05
be a great yeah good training quick one
13:05
be a great yeah good training quick one but I think very very useful yeah
13:08
but I think very very useful yeah
13:08
but I think very very useful yeah absolutely well that was excellent so
13:10
absolutely well that was excellent so
13:10
absolutely well that was excellent so thank you very much for coming again
13:12
thank you very much for coming again
13:12
thank you very much for coming again back to the cloud show and being my
13:14
back to the cloud show and being my
13:14
back to the cloud show and being my first second appearance on the show my
13:17
first second appearance on the show my
13:17
first second appearance on the show my pleasure thank you very much and guests
13:19
pleasure thank you very much and guests
13:19
pleasure thank you very much and guests I'll see you next time on the cloud show
13:26
[Music]
#Machine Learning & Artificial Intelligence