NAB
Virtual humans are an emerging
phenomenon likely to be increasingly used by filmmakers on-screen and
widespread in hospitality and retail scenarios but perhaps alarmingly beginning
to creep into broadcast news and journalism.
article here
The first such virtual humanoid news
anchor in the US, Raxana, was unveiled earlier this year as a lead host of
DeFiance Daily, a program on streaming channel DeFiance.tv.
Marc Scarpa, founder and CEO of parent company
DeFiance Media, is headed to NAB Show New York to
share his experiences in creating Raxana and why he is optimistic for the
future of broadcast news created and presented using AI tools.
Scarpa will provide his insights in a fireside chat
with StoryTech’s Lori H. Schwartz entitled “AI Virtual Humans in Broadcast News.”
During the session, Scarpa “will share the benefits of leveraging AI, along
with human expertise.” Also, learn how DeFiance Media built a bespoke virtual
studio for around-the-clock broadcast news.
“We are not here to destroy
journalism; we are here to empower it,” he tells NAB Amplify. “At the end of
the day, news producers will still need a workflow that involves humans.
Virtual humans are just a device to communicate what those broadcast journalist
stories may be.”
Scarpa predicts that virtual humans
will become pervasive across all sorts of storytelling genres. “You are seeing
it in social media. You will see it in healthcare and education, in other areas
of entertainment, and broadcast journalism will not be immune,” he says.
Part of his argument for AI is simple
cost efficiency: Local news in the US, as in other parts of the world such as
the UK, has been challenged for some time in terms of operating costs. While
local news remains an FCC mandate, its continued existence is in contention
because of changing business models.
Scarpa says he is a big believer in
the value of local news broadcast as a staple of information and culture, but
the economics have to change.
“Broadcasters have to find new ways
to monetize local news. They’ve not veered from the same business model for the
last half century. They have attempted to reduce costs in terms of basic
infrastructure in studios and control rooms and now there’s shift in terms of
talent from an on-camera presence to virtual humans.”
This could be to the benefit of
already stretched on-air talent, suggests Scarpa, since their likeness can be
cloned into a digital representation of them.
“This would allow for the ability for
known personalities to deliver more news packages without having to physically
be there to shoot them and therefore generate more revenue for themselves. It
would be a residual (royalty) module that runs on a blockchain.”
Scarpa is putting his theory into
practice at DeFiance.tv, a television and digital network reporting on the
alternative economy — meaning Web3, AI, VR, and digital asset developments.
Creating and Casting Raxana
During his fireside chat at NAB Show New York,
Scarpa will share lessons learned in the course of creating Raxana. DeFiance
originally worked with an out-of-the-box solution from Israeli developer Hour One but
decided to customize one for their purpose.
“Hour One have fantastic technology,
in particular for business-to-business applications, such as virtual humans for
customer service, but it was not quite the storytelling device we needed for
news. We realized that we needed a virtual human who is representative of our
brand and who could communicate well with our audience.”
Instead, DeFiance used Hour One’s
templates to create a bespoke virtual human based on an existing news anchor
already employed by the network.
“What is most important is you can go
and start with a stock character which is something that we did, but
ultimately, we chose to create our own. That was a huge shift. Our engagement
increased. The resonance with our brand really changed,” he said.
“There is old saying, ‘you have a
face for radio,’ which means something intangible about your ability to connect
with listeners,” Scarpa says. “I believe there is something similar about
people who have an on-camera presence. I am not sure of the science behind it,
but some people are able to communicate better than others in front of a
camera. That art can be replicated, starting by recording motion capture of the
person and delivering their virtual likeness on screen.
“We cast Raxana — based on a real
person — to represent this Eurasian human, which studies predict that most
humans will have this similar genetic mix in future. In essence, Raxana is the
most intelligent journalist that can read a teleprompter in the world, but she
is not going to be doing Barbara Walters interviews anytime soon.”
He continues, “How we use virtual
humans is very rudimentary and basic at this stage. We are not using it to
substitute for investigative journalism or a talk show format. The tech is not
there yet, but it will [be]. I have seen AI virtual companion software that is
outstanding.”
However, DeFinance is using AI across
the news production workflow, not just in presentation.
“We use AI for fact checking and
scripting and research and generation of video. We’re utilizing a variety of AI
tech in our workflow to deliver factual accurate news packages,” Scarpa says.
Other AI Anchors
While the technology is cutting-edge,
Scarpa is not the first to implement it.
In 2018, Chinese News agency Xinhua
News unveiled an AI news anchor. A year later, it launched a female
AI news anchor, “Xin Xiaomeng,” also developed by Chinese tech firm Sogu, whose
voice and image were clones of a reporter named Qu Meng, a news anchor at
Xinhua’s New Media Center.
In April this year, Kuwait News
introduced “Fedha,” an AI news anchor on Twitter, as a test of its potential to
offer new content.
The New Face of News?
Would an artificial news anchor
weaken and destroy any faith we have left in broadcast news as a source of
trusted information?
Scarpa replies by arguing that Walter
Cronkie was the most trusted face of news in history — but that there was still
an organization of broadcast journalists behind the camera.
“There are different types of
broadcasters. Some are investigative, some report sports or the weather. All
the rest of the hosts on our platform are humans who present traditional TV
news shows, doing their own independent research and giving their opinions,” he
says.
“Raxana’s function is very specific.
She is not able to do much, but she is able to do enough for what we want her
deliver and to able to package that news up at scale.
“For example, we can create and
broadcast a news bulletin every two hours and take that package and
auto-translate it into 30 languages. That’s an economy of scale that only AI
can achieve. At the same time, we’re getting across information about the new
economy out to as many people as possible. That is our focus and that is why
Raxana works for us.”
However, live broadcasts are not yet
possible. Scarpa says, “If I have a live camera covering the Senate hearings,
then I would not feel comfortable having that fed into an AI for real-time
commentary as human anchors and journalists do with authority today, but
anything pre-taped can work in that manner.”
That said, Scarpa imagines weather
presentation to evolve to virtual humans “fairly soon” with some forms of
sports commentary to follow — although, again, not where there’s any live
interaction between a virtual sports presenter and an athlete, human news
anchor or pundit.
“We are not there yet with live. I’ve
seen it working and seen you can do it, but it’s not ready for primetime.”
No comments:
Post a Comment