January 1, 2026 • 27 mins
This week we are talking about the biggest moments of 2025 and what all we can look forward to in 2026. Till now we have discussed the trends that shaped politics, sports, entertainment, and health.
Today, we have with us The Indian Express' Karan Mahadik, who shares how the AI landscape evolved during the year, the major success and failures that it experienced and the way it impacted our lives.
Hosted by Niharika NandaProduced by Niharika Nanda and Shashank BhargavaEdited and mixed by Suresh Pawar
Today, we have with us The Indian Express' Karan Mahadik, who shares how the AI landscape evolved during the year, the major success and failures that it experienced and the way it impacted our lives.
Hosted by Niharika NandaProduced by Niharika Nanda and Shashank BhargavaEdited and mixed by Suresh Pawar
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:03):
Hi, I am aherrika Ananda and you are listening to
Three Things the Indian Express News Show. As part of
our ongoing year Ender series, we've been looking back at
some of the key moments of twenty twenty five across
various fields, from cinema and sports to health and politics.
(00:24):
We've discussed the developments that made the biggest difference last year,
and today we are turning our focus to the world
of tech and specifically artificial intelligence, a technology that continues
to be touted as one that will change the world
as we know it. But to understand the extent to
which AI actually delivered on those promises last year, we
(00:45):
are joined by The Indian Express's current Mahalek Current. AI
obviously is now the first thing that comes to mind
when we talk about new technology. So tell us about
some of the trends that dominated the AI landscape in
twenty twenty five.
Speaker 2 (01:00):
Well, first of all, I think thinking back on how
this year began. It began with Donald Trump returning to
the White House, and I think that was quite a
significant moment for the AI companies in the US as
well as the overall air industry, because you know, we
saw these big tech leaders right from Sundar Pichai to
Jeff Bezos to Mark Zuckerberg kind of scrambled to be
(01:22):
in Trump's good graces and try to win him over
by announcing like donations to his inaugural fund. And initially
upon taking office, you know, Trump did take a hardline position.
He wore to break up Meta and impose tariffs on
Apple supply chains and obviously restrict exports of AI chips
from Nvidia and other chip makers to countries like China
(01:44):
and India. But eventually, you know, now that we're nearing
the end of the year, I think most of the
big tech companies working on AI, they've kind of gotten
everything that they've wanted from Trump. I think this year,
as per New York Times, Amazon, Apple, Google, Meta, Nvideer,
Open Ai, Oracle have announced a combined one point four
trillion in spending on domestic data centers and manufacturing projects
(02:07):
in the US. So we saw that twenty twenty five
end up being for air companies based in the US.
They were free of potential regulation after Trump withdrew Biden's
Air Executive Order, and he's put in place, you know,
kind of industry friendly policies.
Speaker 3 (02:24):
We have a tremendous industry where we're leading by a lot.
It's the AI, artificial intelligence. It's a massive industry. We're
leading China, We're leading everybody by a tremendous amount. They can't. Basically,
it's a big part of the economy and there's only
going to be I think one winner here. I don't
(02:45):
know if anybody agrees with that.
Speaker 4 (02:46):
I think most people agree.
Speaker 3 (02:47):
But there's only going to be one winner here, and
that's probably going to be the US or China. And
right now we're winning by a lot at big companies.
Speaker 2 (02:56):
Great Another key trend that emerged was right after this
Trump's integration was the race for AI sovereignty that was
triggered by deep Seak so again in late January, deep
Seak released its AI model, which was an open WATAI
model that performed as well as models developed by open
Air and Google while using fewer resources, and that kind
(03:18):
of triggered a market reaction where there was this sell
off of AI stocks and so on. And essentially what
deep sag did was it kind of made America's lead
over China INAI look smaller than since Chagipity was launched
in twenty twenty two. Another trend was which is kind
of a larger trend outside of US and China, was
(03:40):
a shift in.
Speaker 4 (03:41):
Users search behavior.
Speaker 2 (03:42):
So one of the big looming questions this year was, well,
it wasn't a question really but more of a matter
of fact about how the way people look up information
online is changing.
Speaker 4 (03:52):
Right because earlier you would say google it.
Speaker 2 (03:55):
Now people are literally saying like just chagptiod And obviously
since then, during it iiO Developer conference, Google announced AI Mode,
which is basically different from aioverviews. It kind of replaces
the traditional blue links search results page that we've been
seeing for all these years. It replaces that with the
chatboard that effectively creates a miniature article in response to
(04:18):
your query using various advanced machine learning techniques.
Speaker 5 (04:22):
So AI mode, as you said, yes, it is right
on the home page. It looks like this. That changes
what billions of people see when they open their browser,
which is still the on ramp for the entire internet.
There it is right there. This has major ramifications for
the AI race at large and for the ecosystem that
search has powered for over two decades. So look what
happens when you actually use that AI mode and ask,
(04:45):
say about the Altman Zuckerberg fight, you don't get a
list of news articles. Instead, you get a summary, a
Gemini generated answer, and that matters.
Speaker 2 (04:53):
So there's been this shift in terms of user behavior,
and I think that's also been corroborated by various studies
over the past year. Anset study was conducted by Pure Research,
basically sampling responses from around nine hundred Americans. So this
survey showed that six and ten respondents about fifty eight percent,
conducted at least one Google search in March twenty twenty
(05:13):
five that produced an air generated summary.
Speaker 1 (05:16):
And current If we talk specifically about India and the
development of air language models here, what were some of
the major developments that we saw.
Speaker 2 (05:24):
Well, I think in India too, I think we have
to kind of go back to that Deep Seak market
reaction because when deep Seak released its open weight to
air model, it was also a wake up call for
a lot of Indian developers and Indian researchers and even
the Indian government right and in the days after that,
a lot of questions were rising about where India is
in the global AI areas because historically, if you look
(05:47):
at it, we've been seen as the global back office
for the software industry. And obviously, India has a tech
ecosystem that has mostly evolved with the services first mindset.
You know, we have giants like in forces and tcs
build their success on delivering efficient software. But somewhere I
think invention was neither prioritized or not rewarded enough. And
(06:08):
that's why despite our status as a global tech hub,
we've kind of been seen as lagging far behind US
and China when it comes to developing homegrown AI models right.
And India has also not invested as much in R
and D and institutions and so on. So that was
one key thing that emerged. And just a few days
after the deep Seak thing happened, the Indian government called
(06:31):
for proposals to develop our own foundational AI model under
the India AA Mission. So so far the India AA Mission,
which is like a national level initiative with a more
than ten thousand crore outlay, it was focused on procuring
compute parer and it still is focused on that. But
since then, the Indian government has also been shortlisting a
wide range of startups to develop sovereign lms for India.
Speaker 6 (06:54):
India AI Mission or luksaba k beach may you signed?
Joe Luksa Westruth debates b.
Speaker 4 (07:08):
But iko is co A.
Speaker 6 (07:13):
Joe a AI data set. Jessekiv debate amare your research
Joe Academia unko amare Luksoaki Parliament, Kastruth debate skah translation
(07:37):
but current.
Speaker 1 (07:38):
Even as India develops its own large language models and
a broader AI ecosystem, it has also emerged as a
key market for global AI giants. Would you talk about
what's driving that?
Speaker 4 (07:50):
Right?
Speaker 2 (07:50):
So, the past four five months has seen every major
A company, right from Google to Open Ai to Perplexity,
offering long term free access to their premium AI services
to millions of users in India. Like Perplexity, AI has
tied up with Airtel and eligible Airtail customers get its
Pro subscription for free reliance. Geo and Google a partnered
(08:13):
to offer users a free Google Ai Pro subscription for
eighteen months. OpenAI is also offering an annual low cost
chat gpt go subscription that's worth about four hundred bucks
per month, which is free for one year. And obviously
it makes you wonder why these tech companies are being
so generous, especially at a time when developing AI is
(08:34):
so expensive and the cost of infrastructure investments and training
these AI models. They're spending billions of dollars on it.
So why are they rolling out freebies for India? And
I think based on reports, it could be seen that
one of their strategies is to try and harvest the
rows of multilingual training data from Indian users. Right so,
India is obviously the worldstop consumer of mobile data per
(08:58):
user and its Internet users are to surpass nine hundred billion.
And another problem with developing AI models for specific languages,
especially languages in India, is that Indian language content on
the Internet is very limited and so developers have had
to come up with new unique ways to try and
harvest these data from other sources. And so one of
(09:21):
the ways these companies are doing that is by offering
their models for free for a limited period of time
to India and that sense I think a lot of companies,
from Open Ai to Anthropic have opened new offices and
they've hired personnel to look after the operations here. There's
obviously the mega investments that are being pledged to build
(09:41):
out data centers in India. Just this month Amazon and
Microsoft announced a combined fifty two point five billion investment
plan for India over the coming years. Microsoft committed seventeen
point five billion and Amazon said it would be pumping
in about thirty five billion to help India like strengthen
its AI ecosystem and advance it's like AI driven digitization
(10:03):
as well as you know, create jobs and so on.
Google has also announced a fifteen billion investment to build
an AI data heb and Intel announced a partnership with
Tata so that Tata is its first major customer in
its semiconductor manufacturing plants, which am also about fourteen billion.
So India is not only emerging as a key user market,
(10:24):
but it's also emerging as an AI and cloud infrastructure.
Speaker 1 (10:28):
And in twenty twenty five, AI also became a driver
for several social media trends like the Ghibili style portraits
and meeting my younger self. What do these moments tell
us about how people are engaging with EI today?
Speaker 2 (10:43):
So open I released this new feature called chat GPT
images and it's basically powered by GPT four to Roho
and unlike other image generators because there have been image
generators before that, so these image generators used a diffusion
process to create images, whereas Openey tried to do things
differently and images for chgibt feature takes like a step
(11:06):
by step, top to bottom auto regressive approach for image rendering.
And when this was rolled out, I think it happened organically.
It was not like a deliberate thing. When it was
rolled out, first users started experimenting by generating images in
various cartoon styles like South Park, Rick and Morty and
Simpsons and so on, And then I think one person
(11:27):
on the Internet said like, hey, I made a Studio
Ghibli version of my wife, and the Internet completely lost
its mind. So within a few days, like everyone's social
media feeds was flooded with Studio Ghibli inspired art. But
when this air generated Ghibli art trend kicked off, obviously
there was a lot of backlash as well from Studio
Ghibli fans, who said that the use of air generated
(11:50):
image generators to copy an artist's personal style goes directly
against what visionary co founder of Studio Ghibli stood for.
Others pointed out that it shows how how disposable art
has become thanks to generative AI. But obviously the kind
of fatal flaw. At the heart of all of this
was copyright infringement. Right, so our companies like open ei
(12:11):
training their AI models on copyrighted works, and if so,
does that violate copyright law?
Speaker 7 (12:16):
You may have seen some anime style pictures circulating the
Internet recently, all thanks to a new tool on open
ais GPT.
Speaker 5 (12:24):
Now.
Speaker 7 (12:25):
This one let's use is generate images in studio gibliz
iconic hand drawn visuals similar to those scene in Hayao
Miyazaki's Spirited Away, and the widespread use of the AI
tools also led to questions about creative ownership and potential
copyright violations. A Studio Gibilize founder, the renowned director Hayao Miyazaki,
(12:48):
hasn't yet publicly committed on this recent trend, although he
did previously criticize AI generated animation as quote an insult
to life itself.
Speaker 2 (12:59):
Gosh you can read so after that, I think what
we observed was tech companies like Google in particular, deliberately
trying to ensure that their tools can be used in
such a way. And Google succeeded mostly so it rolled
out Nano Banana, which is its latest image generator, and
what said this imagenerator apart was it could handle highly
(13:21):
specific edits like changing our subject's clothes or locations and
adding another person like a celebrity. So you could edit
your real life photograph and add Sharrukhan next to you
and it.
Speaker 4 (13:32):
Would look really believable, right.
Speaker 2 (13:34):
Obviously, all of these images did come with both a
visible watermark as well as an invisible one called synthid.
Speaker 1 (13:41):
And keeping all of this in mind, do these developments
also raise concerns about more fabricated content and deep fakes,
which are already becoming a serious issue.
Speaker 2 (13:51):
Yeah, I think going forward there will be more concerns
about the growing misuse of synthetically generated information like deep fakes.
The Indian government is also proposed draft rules making the
labeling of air generated content on social media platforms like
YouTube and Instagram mandatory, and these platforms also required to
(14:12):
seek a declaration from users on whether the uploaded content
is synthetically generated information and if they failed to comply,
they could also lose legal immunity they enjoy from third
party content and.
Speaker 1 (14:25):
Karen, another thing that we saw last year was people
using chat Jupity and other AI chatbots to deal with
their mental health issues. People across the world started confiding
in these chatbots. What kind of concerns has this trend raised.
Speaker 2 (14:40):
Yeah, another prevailing trend of this year was the rise
of AI psychosis, which is basically the term used to
describe the experience of when a user loses touch with
reality after using AI chatboards like chad jupity for a
long period of times, and they kind of developed this
kind of attachment to the chatbod even and that in
turn has aked a lot of concerns about how AI
(15:02):
chatbots and AI companion bots affect users' mental health, particularly
teen users. So recently, Opena's own study said that around
zero point zero seven percent of Chagibility users showed signs
of mental health emergencies like mania, psychosis, and suicidal thoughts.
And while that might seem like a small percentage, Chagability
(15:26):
is weekly active user basis seven hundred million, so it
actually translates to more than a million. And besides that,
you know, Openia is obviously facing a lot of lawsuits
which alleged that teen users are allegedly diet by suicide
after having prolonged information conversations with a chatbots. So and
all of this has brought opena and a lot of
(15:46):
scrutiny from like policymakers, educators, and child safety advocates.
Speaker 1 (15:51):
And looking back at twenty twenty five, what were some
of the major developments in AI and were there any
surprises along the way right?
Speaker 2 (15:59):
Well, for me, I think the most unexpected and probably
the one that I'm most excited about as well is
Google's comeback with Gemini three. So when open i introduced
chargpt almost three years ago, it was kind of seen
as this starting pistol in the AI race for all
tech companies, but for Google in particular, because they wrote
the Google researchers invented the concept of the transformer, which
(16:22):
is the underlying tech that has made elms possible. And
at that point Google was caught on the back foot.
They went into a like a code red situation to
try and compete with open ai, and since then, the
surch giant has come a long way. They've tried to
disrupt themselves by kind of breaking down the internal silos,
streamlining leadership, consolidating their AI projects, and just overhauling their
(16:45):
entire strategy to AI development. Even Sergebrin, one of Google's
co founders, came out of partial retirement and has since
taken on like a day to day role at the companies,
And obviously Google is also so this is.
Speaker 4 (16:57):
Kind of like a two front batter.
Speaker 2 (16:59):
On one hand, it is competing with OpenAI as to
who has the better LLM, better AI model. On the
other end, it is also manufacturing its own chips to
kind of reduce its dependence on Invidia's GPUs, and I
think Gemini three was also seen as a milestone on
that front because it was trained entirely on Google's TPUs,
which is its custom built tensor processing units, rather than
(17:21):
in Video's GPUs. That also triggered a market reaction which
led to like shares declining of Nvideo, but obviously not
as big as the deep seak reaction, but it was
quite substantial and made a lot of people pay attention
to it. But again, I think it's also important to
note that a lot of people like SATNDLA have argued
that AI model development and cloud computing, they're not likely
(17:44):
to be winner takes all scenarios, right, So yeah, that's
one point to be noted.
Speaker 1 (17:49):
Now that we've talked about the developments, what about the
failures with AI still being very much a work in
progress and with new models coming out all the time,
what didn't work out this?
Speaker 2 (18:00):
Yeah, yeah, there have been many failures. Well I would
actually call them the biggest fails of twenty twenty five.
The first and foremost would be Openized GPT five launch.
I think Openey itself said, you know, it screwed up
the rollout of the AI model GPT five. In fact,
the tone and behavior of chagipt has been kind of
an ongoing issue this year. So what happened was when
(18:23):
the release of GPT five was widely anticipated and was
built up by Openea itself. You know, Sam Altman posted
this image of the death Star on the horizon like
a few days before the AI model was rolled out,
and so there's all of these rumors circulating on social
media and a lot of users expected a lot, But
when the model was relieved, I think a lot of
(18:44):
them found it underwhelming in terms of its responses. A
lot of them complained that, you know, the responses were
overly agreeable, that they were sycophantic. And then just a
few days after the model was officially launched, open I
rolled out another behavioral update which right to make GPT
fire's responses kind of warmer and friendlier after users complained
(19:05):
that was very cold and less friendly. It made a
series of tweaks and behavioral adjustments to the model following
its debut. So that was one failure that comes to mind.
Another was probably City and how generative. AIS kind of
failed to boost the AI powered voice assistance like City
and Alexa plus. For City, I think you know Apple
(19:26):
to a large extent, there is this perception that in
both Silicon Valley on on Wall Street that Apple has
fallen behind its peers in AI. Apple launched Apple Intelligence
last year but amid a lot of expectations. But in
the initial days a lot of people hyped the AI
system as like the next big thing, but ultimately, I
think twenty twenty five users realized that Apple Intelligence falls
(19:50):
short of those expectations. Both critics and Apple enthusiasts were
disappointed with Apple Intelligence. A major part of that problem
is also City was the expect to undergo like a
aicentric makeover, but due to various issues, the overhaul was
taking longer than expected, and an Apple has now delayed
the release of that revamped AI makeover City until twenty
(20:13):
twenty six. Another fail of twenty twenty five in Ai
was Grog.
Speaker 8 (20:19):
Why was Elon Musk's croc ai chatbot suddenly spewing out
anti Semitic tropes? Well, a few weeks ago Elon Musk
announce so they would be completely retraining Grog. Musk had
expressed disappointment with how Grok had responded to some users
claiming that it was quote parroting legacy media. A few
days ago. He said that users should start noticing a change,
and users did notice that change, and in some cases
it raised some alarms. In one case, Groc praised Adolf Hitler,
(20:42):
saying that Hitler would be the one who would be
able to decisively deal with what it said was anti
white hate. And another instance, when a user asked who
controls the government, Grok seemingly implied that Jews, who make
up just about two percent of the US population, were
over represented in positions.
Speaker 1 (20:57):
Of power and influence.
Speaker 2 (20:59):
I asked Rock where some of these Grock showed what
runaway AAR models can do. It it's unhinged responses sparked
like a full borne controversy in India, and mostly because
you know, Grock is not like other a chat bots.
It's not designed like that it's not as locked down
as other chatbots when it comes to like politically sensitive topics.
(21:19):
It actually offers users in unhinged mode for premium paid users,
that is, and Grock has also been openly advertised by
Musk as like an anti WO alternative. In an interview
last year, you said that Grog happened mainly because of
fears that existing AAR models had left being biased baked
into their training data sets. And there have been so
(21:41):
many instances this year of Grock going off the rails.
Grog gone wild, It praised out of Hitler, it even
turned against its own company owner, Elon Musk, And I
think a lot of these issues also point to like
a more fundamental problem with lms, which is, you know,
it's hard to predict their behavior and the responses.
Speaker 1 (22:01):
Besides AI models like Roc going off the rails, there's
also this constant narrative that AI would end up taking
away a lot of jobs given the pace at which
is developing. Is that developed fear.
Speaker 4 (22:13):
I think it's a valid fear.
Speaker 2 (22:15):
While the uptake of AI was so in twenty twenty four,
AI was widely recognized as contributing to layoffs, but most
of the companies framed the job cuts as a response
to broader economic uncertainty, and very few firms are willing
to explicitly link job cuts to AI or acknowledging that
they were replacing humans with AI tools. But in twenty
(22:38):
twenty five that changed. This year was very different in
that sense. I think a lot of major tech companies,
right from Amazon to Microsoft, announced thousands of job cuts,
and more importantly, they cited AI or AI adoption as
a key factor of those job cuts. I think we'll
have to also watch out for whether AI generated music
(22:59):
and ads whether they can become mainstream. I think in
twenty twenty five we've already seen the proliferation of AI slop.
But besides that, we've also seen a lot of AA
generated musicians find success. You know, a lot of tools
have been released to generate AI podcasts. There have been
tools that have been used to create air generated videos
(23:20):
that feature dead celebrities. In September, an AI startup released
a very realistic appearing AI generated actress called Tilly Norvewood,
and that kind of spark concern in Hollywood over job replacement.
Speaker 7 (23:35):
I'm Tilly Norwood, the world's first AI actor come and work.
Speaker 2 (23:39):
With me, so we Obviously a lot of filmmakers have
come out against the use of AI in films, people
like Gamo, Delta Ro and James Cameron. And on the
other hand, you know, you have companies like Disney reaching
agreements with open Ai bickalarly allowing open ai to use
Disney owned characters in it AA generated videos. So that'll
(24:01):
be another thing to watch out for.
Speaker 1 (24:03):
And Kuren, Now that twenty twenty six is here, what
are some of the biggest expectations from AI going forward?
Speaker 2 (24:10):
I think twenty twenty six is bound to be another
significant year for AID. I will be hosting the India
AI Impact Summit from fifteen to twenty eight and it's
supposed to be one of the biggest events AI related
events in the global South.
Speaker 4 (24:25):
About heads of.
Speaker 2 (24:26):
Government from around fifteen to twenty countries and several global
figures from A and tech are expected to participate. It
will be inaugurated by PMO. The India sent out invitations
to about one hundred and forty countries so far, and
the conferences is being held to generate recommendations for how
to govern AI long term and also come up with
(24:49):
more immediate regulations in terms of AI, but most specifically,
I think there are a few things that one should
watch out for. For instance, AI could pose a threat
to the creata A lot of creators already under pressure
from platforms because they're not being paid very well, and
that's pushing them towards brand deals and as a result,
they're generating more inauthentic content or content that not necessarily
(25:13):
resonates with their audience. Add to that, AI, I mean
a lot of big tech companies are viewing AI generated
content as the future of social media. You know, you
had people like Mark zuckerberg'say and Earnings called that Meta
plans to add another huge corpus of content, and it
plans to kind of refine its recommendation systems to make
(25:33):
it easier for users to create and remix AI into
their work. And finally, another thing that is very widely
expected in twenty twenty six is the platform shift that
so many tech industry leaders have been talking about. While
we have AI powered devices, and while smartphones have been
integrated with a lot of AI features and so on,
(25:54):
including laptops as well, I think no single companies yet
perfected and entirely AI focused hardware. You know, OpenAI partnering
with Johnny Ive who was a former Apple designer basically
in charge of how the iPhone looks today. And so
while no one as yet has been successful and in
(26:15):
coming up with any kind of gadget or hardware that
is completely built around an LLM, it's quite interesting to see,
you know, how this partnership will pan out.
Speaker 1 (26:27):
You were listening to Three Things by the Nail Express.
Today's show was edited and mixed by Srish Bavar and
produced by Shashank Pargev and Nina Harrikananda. If you like
the show, then do subscribe to us wherever you get
your podcasts. You can also recommend the show to someone
you think will like it, share it with a friend
or someone in a family. It's the best way for
people to get to know about us. You can also
(26:49):
tweet us at Express Audio and write to us at
podcasts at Indian Express dot com.
Hi, I am aherrika Ananda and you are listening to
Three Things the Indian Express News Show. As part of
our ongoing year Ender series, we've been looking back at
some of the key moments of twenty twenty five across
various fields, from cinema and sports to health and politics.
(00:24):
We've discussed the developments that made the biggest difference last year,
and today we are turning our focus to the world
of tech and specifically artificial intelligence, a technology that continues
to be touted as one that will change the world
as we know it. But to understand the extent to
which AI actually delivered on those promises last year, we
(00:45):
are joined by The Indian Express's current Mahalek Current. AI
obviously is now the first thing that comes to mind
when we talk about new technology. So tell us about
some of the trends that dominated the AI landscape in
twenty twenty five.
Speaker 2 (01:00):
Well, first of all, I think thinking back on how
this year began. It began with Donald Trump returning to
the White House, and I think that was quite a
significant moment for the AI companies in the US as
well as the overall air industry, because you know, we
saw these big tech leaders right from Sundar Pichai to
Jeff Bezos to Mark Zuckerberg kind of scrambled to be
(01:22):
in Trump's good graces and try to win him over
by announcing like donations to his inaugural fund. And initially
upon taking office, you know, Trump did take a hardline position.
He wore to break up Meta and impose tariffs on
Apple supply chains and obviously restrict exports of AI chips
from Nvidia and other chip makers to countries like China
(01:44):
and India. But eventually, you know, now that we're nearing
the end of the year, I think most of the
big tech companies working on AI, they've kind of gotten
everything that they've wanted from Trump. I think this year,
as per New York Times, Amazon, Apple, Google, Meta, Nvideer,
Open Ai, Oracle have announced a combined one point four
trillion in spending on domestic data centers and manufacturing projects
(02:07):
in the US. So we saw that twenty twenty five
end up being for air companies based in the US.
They were free of potential regulation after Trump withdrew Biden's
Air Executive Order, and he's put in place, you know,
kind of industry friendly policies.
Speaker 3 (02:24):
We have a tremendous industry where we're leading by a lot.
It's the AI, artificial intelligence. It's a massive industry. We're
leading China, We're leading everybody by a tremendous amount. They can't. Basically,
it's a big part of the economy and there's only
going to be I think one winner here. I don't
(02:45):
know if anybody agrees with that.
Speaker 4 (02:46):
I think most people agree.
Speaker 3 (02:47):
But there's only going to be one winner here, and
that's probably going to be the US or China. And
right now we're winning by a lot at big companies.
Speaker 2 (02:56):
Great Another key trend that emerged was right after this
Trump's integration was the race for AI sovereignty that was
triggered by deep Seak so again in late January, deep
Seak released its AI model, which was an open WATAI
model that performed as well as models developed by open
Air and Google while using fewer resources, and that kind
(03:18):
of triggered a market reaction where there was this sell
off of AI stocks and so on. And essentially what
deep sag did was it kind of made America's lead
over China INAI look smaller than since Chagipity was launched
in twenty twenty two. Another trend was which is kind
of a larger trend outside of US and China, was
(03:40):
a shift in.
Speaker 4 (03:41):
Users search behavior.
Speaker 2 (03:42):
So one of the big looming questions this year was, well,
it wasn't a question really but more of a matter
of fact about how the way people look up information
online is changing.
Speaker 4 (03:52):
Right because earlier you would say google it.
Speaker 2 (03:55):
Now people are literally saying like just chagptiod And obviously
since then, during it iiO Developer conference, Google announced AI Mode,
which is basically different from aioverviews. It kind of replaces
the traditional blue links search results page that we've been
seeing for all these years. It replaces that with the
chatboard that effectively creates a miniature article in response to
(04:18):
your query using various advanced machine learning techniques.
Speaker 5 (04:22):
So AI mode, as you said, yes, it is right
on the home page. It looks like this. That changes
what billions of people see when they open their browser,
which is still the on ramp for the entire internet.
There it is right there. This has major ramifications for
the AI race at large and for the ecosystem that
search has powered for over two decades. So look what
happens when you actually use that AI mode and ask,
(04:45):
say about the Altman Zuckerberg fight, you don't get a
list of news articles. Instead, you get a summary, a
Gemini generated answer, and that matters.
Speaker 2 (04:53):
So there's been this shift in terms of user behavior,
and I think that's also been corroborated by various studies
over the past year. Anset study was conducted by Pure Research,
basically sampling responses from around nine hundred Americans. So this
survey showed that six and ten respondents about fifty eight percent,
conducted at least one Google search in March twenty twenty
(05:13):
five that produced an air generated summary.
Speaker 1 (05:16):
And current If we talk specifically about India and the
development of air language models here, what were some of
the major developments that we saw.
Speaker 2 (05:24):
Well, I think in India too, I think we have
to kind of go back to that Deep Seak market
reaction because when deep Seak released its open weight to
air model, it was also a wake up call for
a lot of Indian developers and Indian researchers and even
the Indian government right and in the days after that,
a lot of questions were rising about where India is
in the global AI areas because historically, if you look
(05:47):
at it, we've been seen as the global back office
for the software industry. And obviously, India has a tech
ecosystem that has mostly evolved with the services first mindset.
You know, we have giants like in forces and tcs
build their success on delivering efficient software. But somewhere I
think invention was neither prioritized or not rewarded enough. And
(06:08):
that's why despite our status as a global tech hub,
we've kind of been seen as lagging far behind US
and China when it comes to developing homegrown AI models right.
And India has also not invested as much in R
and D and institutions and so on. So that was
one key thing that emerged. And just a few days
after the deep Seak thing happened, the Indian government called
(06:31):
for proposals to develop our own foundational AI model under
the India AA Mission. So so far the India AA Mission,
which is like a national level initiative with a more
than ten thousand crore outlay, it was focused on procuring
compute parer and it still is focused on that. But
since then, the Indian government has also been shortlisting a
wide range of startups to develop sovereign lms for India.
Speaker 6 (06:54):
India AI Mission or luksaba k beach may you signed?
Joe Luksa Westruth debates b.
Speaker 4 (07:08):
But iko is co A.
Speaker 6 (07:13):
Joe a AI data set. Jessekiv debate amare your research
Joe Academia unko amare Luksoaki Parliament, Kastruth debate skah translation
(07:37):
but current.
Speaker 1 (07:38):
Even as India develops its own large language models and
a broader AI ecosystem, it has also emerged as a
key market for global AI giants. Would you talk about
what's driving that?
Speaker 4 (07:50):
Right?
Speaker 2 (07:50):
So, the past four five months has seen every major
A company, right from Google to Open Ai to Perplexity,
offering long term free access to their premium AI services
to millions of users in India. Like Perplexity, AI has
tied up with Airtel and eligible Airtail customers get its
Pro subscription for free reliance. Geo and Google a partnered
(08:13):
to offer users a free Google Ai Pro subscription for
eighteen months. OpenAI is also offering an annual low cost
chat gpt go subscription that's worth about four hundred bucks
per month, which is free for one year. And obviously
it makes you wonder why these tech companies are being
so generous, especially at a time when developing AI is
(08:34):
so expensive and the cost of infrastructure investments and training
these AI models. They're spending billions of dollars on it.
So why are they rolling out freebies for India? And
I think based on reports, it could be seen that
one of their strategies is to try and harvest the
rows of multilingual training data from Indian users. Right so,
India is obviously the worldstop consumer of mobile data per
(08:58):
user and its Internet users are to surpass nine hundred billion.
And another problem with developing AI models for specific languages,
especially languages in India, is that Indian language content on
the Internet is very limited and so developers have had
to come up with new unique ways to try and
harvest these data from other sources. And so one of
(09:21):
the ways these companies are doing that is by offering
their models for free for a limited period of time
to India and that sense I think a lot of companies,
from Open Ai to Anthropic have opened new offices and
they've hired personnel to look after the operations here. There's
obviously the mega investments that are being pledged to build
(09:41):
out data centers in India. Just this month Amazon and
Microsoft announced a combined fifty two point five billion investment
plan for India over the coming years. Microsoft committed seventeen
point five billion and Amazon said it would be pumping
in about thirty five billion to help India like strengthen
its AI ecosystem and advance it's like AI driven digitization
(10:03):
as well as you know, create jobs and so on.
Google has also announced a fifteen billion investment to build
an AI data heb and Intel announced a partnership with
Tata so that Tata is its first major customer in
its semiconductor manufacturing plants, which am also about fourteen billion.
So India is not only emerging as a key user market,
(10:24):
but it's also emerging as an AI and cloud infrastructure.
Speaker 1 (10:28):
And in twenty twenty five, AI also became a driver
for several social media trends like the Ghibili style portraits
and meeting my younger self. What do these moments tell
us about how people are engaging with EI today?
Speaker 2 (10:43):
So open I released this new feature called chat GPT
images and it's basically powered by GPT four to Roho
and unlike other image generators because there have been image
generators before that, so these image generators used a diffusion
process to create images, whereas Openey tried to do things
differently and images for chgibt feature takes like a step
(11:06):
by step, top to bottom auto regressive approach for image rendering.
And when this was rolled out, I think it happened organically.
It was not like a deliberate thing. When it was
rolled out, first users started experimenting by generating images in
various cartoon styles like South Park, Rick and Morty and
Simpsons and so on, And then I think one person
(11:27):
on the Internet said like, hey, I made a Studio
Ghibli version of my wife, and the Internet completely lost
its mind. So within a few days, like everyone's social
media feeds was flooded with Studio Ghibli inspired art. But
when this air generated Ghibli art trend kicked off, obviously
there was a lot of backlash as well from Studio
Ghibli fans, who said that the use of air generated
(11:50):
image generators to copy an artist's personal style goes directly
against what visionary co founder of Studio Ghibli stood for.
Others pointed out that it shows how how disposable art
has become thanks to generative AI. But obviously the kind
of fatal flaw. At the heart of all of this
was copyright infringement. Right, so our companies like open ei
(12:11):
training their AI models on copyrighted works, and if so,
does that violate copyright law?
Speaker 7 (12:16):
You may have seen some anime style pictures circulating the
Internet recently, all thanks to a new tool on open
ais GPT.
Speaker 5 (12:24):
Now.
Speaker 7 (12:25):
This one let's use is generate images in studio gibliz
iconic hand drawn visuals similar to those scene in Hayao
Miyazaki's Spirited Away, and the widespread use of the AI
tools also led to questions about creative ownership and potential
copyright violations. A Studio Gibilize founder, the renowned director Hayao Miyazaki,
(12:48):
hasn't yet publicly committed on this recent trend, although he
did previously criticize AI generated animation as quote an insult
to life itself.
Speaker 2 (12:59):
Gosh you can read so after that, I think what
we observed was tech companies like Google in particular, deliberately
trying to ensure that their tools can be used in
such a way. And Google succeeded mostly so it rolled
out Nano Banana, which is its latest image generator, and
what said this imagenerator apart was it could handle highly
(13:21):
specific edits like changing our subject's clothes or locations and
adding another person like a celebrity. So you could edit
your real life photograph and add Sharrukhan next to you
and it.
Speaker 4 (13:32):
Would look really believable, right.
Speaker 2 (13:34):
Obviously, all of these images did come with both a
visible watermark as well as an invisible one called synthid.
Speaker 1 (13:41):
And keeping all of this in mind, do these developments
also raise concerns about more fabricated content and deep fakes,
which are already becoming a serious issue.
Speaker 2 (13:51):
Yeah, I think going forward there will be more concerns
about the growing misuse of synthetically generated information like deep fakes.
The Indian government is also proposed draft rules making the
labeling of air generated content on social media platforms like
YouTube and Instagram mandatory, and these platforms also required to
(14:12):
seek a declaration from users on whether the uploaded content
is synthetically generated information and if they failed to comply,
they could also lose legal immunity they enjoy from third
party content and.
Speaker 1 (14:25):
Karen, another thing that we saw last year was people
using chat Jupity and other AI chatbots to deal with
their mental health issues. People across the world started confiding
in these chatbots. What kind of concerns has this trend raised.
Speaker 2 (14:40):
Yeah, another prevailing trend of this year was the rise
of AI psychosis, which is basically the term used to
describe the experience of when a user loses touch with
reality after using AI chatboards like chad jupity for a
long period of times, and they kind of developed this
kind of attachment to the chatbod even and that in
turn has aked a lot of concerns about how AI
(15:02):
chatbots and AI companion bots affect users' mental health, particularly
teen users. So recently, Opena's own study said that around
zero point zero seven percent of Chagibility users showed signs
of mental health emergencies like mania, psychosis, and suicidal thoughts.
And while that might seem like a small percentage, Chagability
(15:26):
is weekly active user basis seven hundred million, so it
actually translates to more than a million. And besides that,
you know, Openia is obviously facing a lot of lawsuits
which alleged that teen users are allegedly diet by suicide
after having prolonged information conversations with a chatbots. So and
all of this has brought opena and a lot of
(15:46):
scrutiny from like policymakers, educators, and child safety advocates.
Speaker 1 (15:51):
And looking back at twenty twenty five, what were some
of the major developments in AI and were there any
surprises along the way right?
Speaker 2 (15:59):
Well, for me, I think the most unexpected and probably
the one that I'm most excited about as well is
Google's comeback with Gemini three. So when open i introduced
chargpt almost three years ago, it was kind of seen
as this starting pistol in the AI race for all
tech companies, but for Google in particular, because they wrote
the Google researchers invented the concept of the transformer, which
(16:22):
is the underlying tech that has made elms possible. And
at that point Google was caught on the back foot.
They went into a like a code red situation to
try and compete with open ai, and since then, the
surch giant has come a long way. They've tried to
disrupt themselves by kind of breaking down the internal silos,
streamlining leadership, consolidating their AI projects, and just overhauling their
(16:45):
entire strategy to AI development. Even Sergebrin, one of Google's
co founders, came out of partial retirement and has since
taken on like a day to day role at the companies,
And obviously Google is also so this is.
Speaker 4 (16:57):
Kind of like a two front batter.
Speaker 2 (16:59):
On one hand, it is competing with OpenAI as to
who has the better LLM, better AI model. On the
other end, it is also manufacturing its own chips to
kind of reduce its dependence on Invidia's GPUs, and I
think Gemini three was also seen as a milestone on
that front because it was trained entirely on Google's TPUs,
which is its custom built tensor processing units, rather than
(17:21):
in Video's GPUs. That also triggered a market reaction which
led to like shares declining of Nvideo, but obviously not
as big as the deep seak reaction, but it was
quite substantial and made a lot of people pay attention
to it. But again, I think it's also important to
note that a lot of people like SATNDLA have argued
that AI model development and cloud computing, they're not likely
(17:44):
to be winner takes all scenarios, right, So yeah, that's
one point to be noted.
Speaker 1 (17:49):
Now that we've talked about the developments, what about the
failures with AI still being very much a work in
progress and with new models coming out all the time,
what didn't work out this?
Speaker 2 (18:00):
Yeah, yeah, there have been many failures. Well I would
actually call them the biggest fails of twenty twenty five.
The first and foremost would be Openized GPT five launch.
I think Openey itself said, you know, it screwed up
the rollout of the AI model GPT five. In fact,
the tone and behavior of chagipt has been kind of
an ongoing issue this year. So what happened was when
(18:23):
the release of GPT five was widely anticipated and was
built up by Openea itself. You know, Sam Altman posted
this image of the death Star on the horizon like
a few days before the AI model was rolled out,
and so there's all of these rumors circulating on social
media and a lot of users expected a lot, But
when the model was relieved, I think a lot of
(18:44):
them found it underwhelming in terms of its responses. A
lot of them complained that, you know, the responses were
overly agreeable, that they were sycophantic. And then just a
few days after the model was officially launched, open I
rolled out another behavioral update which right to make GPT
fire's responses kind of warmer and friendlier after users complained
(19:05):
that was very cold and less friendly. It made a
series of tweaks and behavioral adjustments to the model following
its debut. So that was one failure that comes to mind.
Another was probably City and how generative. AIS kind of
failed to boost the AI powered voice assistance like City
and Alexa plus. For City, I think you know Apple
(19:26):
to a large extent, there is this perception that in
both Silicon Valley on on Wall Street that Apple has
fallen behind its peers in AI. Apple launched Apple Intelligence
last year but amid a lot of expectations. But in
the initial days a lot of people hyped the AI
system as like the next big thing, but ultimately, I
think twenty twenty five users realized that Apple Intelligence falls
(19:50):
short of those expectations. Both critics and Apple enthusiasts were
disappointed with Apple Intelligence. A major part of that problem
is also City was the expect to undergo like a
aicentric makeover, but due to various issues, the overhaul was
taking longer than expected, and an Apple has now delayed
the release of that revamped AI makeover City until twenty
(20:13):
twenty six. Another fail of twenty twenty five in Ai
was Grog.
Speaker 8 (20:19):
Why was Elon Musk's croc ai chatbot suddenly spewing out
anti Semitic tropes? Well, a few weeks ago Elon Musk
announce so they would be completely retraining Grog. Musk had
expressed disappointment with how Grok had responded to some users
claiming that it was quote parroting legacy media. A few
days ago. He said that users should start noticing a change,
and users did notice that change, and in some cases
it raised some alarms. In one case, Groc praised Adolf Hitler,
(20:42):
saying that Hitler would be the one who would be
able to decisively deal with what it said was anti
white hate. And another instance, when a user asked who
controls the government, Grok seemingly implied that Jews, who make
up just about two percent of the US population, were
over represented in positions.
Speaker 1 (20:57):
Of power and influence.
Speaker 2 (20:59):
I asked Rock where some of these Grock showed what
runaway AAR models can do. It it's unhinged responses sparked
like a full borne controversy in India, and mostly because
you know, Grock is not like other a chat bots.
It's not designed like that it's not as locked down
as other chatbots when it comes to like politically sensitive topics.
(21:19):
It actually offers users in unhinged mode for premium paid users,
that is, and Grock has also been openly advertised by
Musk as like an anti WO alternative. In an interview
last year, you said that Grog happened mainly because of
fears that existing AAR models had left being biased baked
into their training data sets. And there have been so
(21:41):
many instances this year of Grock going off the rails.
Grog gone wild, It praised out of Hitler, it even
turned against its own company owner, Elon Musk, And I
think a lot of these issues also point to like
a more fundamental problem with lms, which is, you know,
it's hard to predict their behavior and the responses.
Speaker 1 (22:01):
Besides AI models like Roc going off the rails, there's
also this constant narrative that AI would end up taking
away a lot of jobs given the pace at which
is developing. Is that developed fear.
Speaker 4 (22:13):
I think it's a valid fear.
Speaker 2 (22:15):
While the uptake of AI was so in twenty twenty four,
AI was widely recognized as contributing to layoffs, but most
of the companies framed the job cuts as a response
to broader economic uncertainty, and very few firms are willing
to explicitly link job cuts to AI or acknowledging that
they were replacing humans with AI tools. But in twenty
(22:38):
twenty five that changed. This year was very different in
that sense. I think a lot of major tech companies,
right from Amazon to Microsoft, announced thousands of job cuts,
and more importantly, they cited AI or AI adoption as
a key factor of those job cuts. I think we'll
have to also watch out for whether AI generated music
(22:59):
and ads whether they can become mainstream. I think in
twenty twenty five we've already seen the proliferation of AI slop.
But besides that, we've also seen a lot of AA
generated musicians find success. You know, a lot of tools
have been released to generate AI podcasts. There have been
tools that have been used to create air generated videos
(23:20):
that feature dead celebrities. In September, an AI startup released
a very realistic appearing AI generated actress called Tilly Norvewood,
and that kind of spark concern in Hollywood over job replacement.
Speaker 7 (23:35):
I'm Tilly Norwood, the world's first AI actor come and work.
Speaker 2 (23:39):
With me, so we Obviously a lot of filmmakers have
come out against the use of AI in films, people
like Gamo, Delta Ro and James Cameron. And on the
other hand, you know, you have companies like Disney reaching
agreements with open Ai bickalarly allowing open ai to use
Disney owned characters in it AA generated videos. So that'll
(24:01):
be another thing to watch out for.
Speaker 1 (24:03):
And Kuren, Now that twenty twenty six is here, what
are some of the biggest expectations from AI going forward?
Speaker 2 (24:10):
I think twenty twenty six is bound to be another
significant year for AID. I will be hosting the India
AI Impact Summit from fifteen to twenty eight and it's
supposed to be one of the biggest events AI related
events in the global South.
Speaker 4 (24:25):
About heads of.
Speaker 2 (24:26):
Government from around fifteen to twenty countries and several global
figures from A and tech are expected to participate. It
will be inaugurated by PMO. The India sent out invitations
to about one hundred and forty countries so far, and
the conferences is being held to generate recommendations for how
to govern AI long term and also come up with
(24:49):
more immediate regulations in terms of AI, but most specifically,
I think there are a few things that one should
watch out for. For instance, AI could pose a threat
to the creata A lot of creators already under pressure
from platforms because they're not being paid very well, and
that's pushing them towards brand deals and as a result,
they're generating more inauthentic content or content that not necessarily
(25:13):
resonates with their audience. Add to that, AI, I mean
a lot of big tech companies are viewing AI generated
content as the future of social media. You know, you
had people like Mark zuckerberg'say and Earnings called that Meta
plans to add another huge corpus of content, and it
plans to kind of refine its recommendation systems to make
(25:33):
it easier for users to create and remix AI into
their work. And finally, another thing that is very widely
expected in twenty twenty six is the platform shift that
so many tech industry leaders have been talking about. While
we have AI powered devices, and while smartphones have been
integrated with a lot of AI features and so on,
(25:54):
including laptops as well, I think no single companies yet
perfected and entirely AI focused hardware. You know, OpenAI partnering
with Johnny Ive who was a former Apple designer basically
in charge of how the iPhone looks today. And so
while no one as yet has been successful and in
(26:15):
coming up with any kind of gadget or hardware that
is completely built around an LLM, it's quite interesting to see,
you know, how this partnership will pan out.
Speaker 1 (26:27):
You were listening to Three Things by the Nail Express.
Today's show was edited and mixed by Srish Bavar and
produced by Shashank Pargev and Nina Harrikananda. If you like
the show, then do subscribe to us wherever you get
your podcasts. You can also recommend the show to someone
you think will like it, share it with a friend
or someone in a family. It's the best way for
people to get to know about us. You can also
(26:49):
tweet us at Express Audio and write to us at
podcasts at Indian Express dot com.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
The Burden
The Burden is a documentary series that takes listeners into the hidden places where justice is done (and undone). It dives deep into the lives of heroes and villains. And it focuses a spotlight on those who triumph even when the odds are against them. Season 5 - The Burden: Death & Deceit in Alliance On April Fools Day 1999, 26-year-old Yvonne Layne was found murdered in her Alliance, Ohio home. David Thorne, her ex-boyfriend and father of one of her children, was instantly a suspect. Another young man admitted to the murder, and David breathed a sigh of relief, until the confessed murderer fingered David; “He paid me to do it.” David was sentenced to life without parole. Two decades later, Pulitzer winner and podcast host, Maggie Freleng (Bone Valley Season 3: Graves County, Wrongful Conviction, Suave) launched a “live” investigation into David's conviction alongside Jason Baldwin (himself wrongfully convicted as a member of the West Memphis Three). Maggie had come to believe that the entire investigation of David was botched by the tiny local police department, or worse, covered up the real killer. Was Maggie correct? Was David’s claim of innocence credible? In Death and Deceit in Alliance, Maggie recounts the case that launched her career, and ultimately, “broke” her.” The results will shock the listener and reduce Maggie to tears and self-doubt. This is not your typical wrongful conviction story. In fact, it turns the genre on its head. It asks the question: What if our champions are foolish? Season 4 - The Burden: Get the Money and Run “Trying to murder my father, this was the thing that put me on the path.” That’s Joe Loya and that path was bank robbery. Bank, bank, bank, bank, bank. In season 4 of The Burden: Get the Money and Run, we hear from Joe who was once the most prolific bank robber in Southern California, and beyond. He used disguises, body doubles, proxies. He leaped over counters, grabbed the money and ran. Even as the FBI was closing in. It was a showdown between a daring bank robber, and a patient FBI agent. Joe was no ordinary bank robber. He was bright, articulate, charismatic, and driven by a dark rage that he summoned up at will. In seven episodes, Joe tells all: the what, the how… and the why. Including why he tried to murder his father. Season 3 - The Burden: Avenger Miriam Lewin is one of Argentina’s leading journalists today. At 19 years old, she was kidnapped off the streets of Buenos Aires for her political activism and thrown into a concentration camp. Thousands of her fellow inmates were executed, tossed alive from a cargo plane into the ocean. Miriam, along with a handful of others, will survive the camp. Then as a journalist, she will wage a decades long campaign to bring her tormentors to justice. Avenger is about one woman’s triumphant battle against unbelievable odds to survive torture, claim justice for the crimes done against her and others like her, and change the future of her country. Season 2 - The Burden: Empire on Blood Empire on Blood is set in the Bronx, NY, in the early 90s, when two young drug dealers ruled an intersection known as “The Corner on Blood.” The boss, Calvin Buari, lived large. He and a protege swore they would build an empire on blood. Then the relationship frayed and the protege accused Calvin of a double homicide which he claimed he didn’t do. But did he? Award-winning journalist Steve Fishman spent seven years to answer that question. This is the story of one man’s last chance to overturn his life sentence. He may prevail, but someone’s gotta pay. The Burden: Empire on Blood is the director’s cut of the true crime classic which reached #1 on the charts when it was first released half a dozen years ago. Season 1 - The Burden In the 1990s, Detective Louis N. Scarcella was legendary. In a city overrun by violent crime, he cracked the toughest cases and put away the worst criminals. “The Hulk” was his nickname. Then the story changed. Scarcella ran into a group of convicted murderers who all say they are innocent. They turned themselves into jailhouse-lawyers and in prison founded a lway firm. When they realized Scarcella helped put many of them away, they set their sights on taking him down. And with the help of a NY Times reporter they have a chance. For years, Scarcella insisted he did nothing wrong. But that’s all he’d say. Until we tracked Scarcella to a sauna in a Russian bathhouse, where he started to talk..and talk and talk. “The guilty have gone free,” he whispered. And then agreed to take us into the belly of the beast. Welcome to The Burden.