NAB
article here
On New Year’s Eve, OpenAI president and co- founder Greg Brockman (@gdb)
tweeted: “Prediction: 2023 will make 2022 look like a sleepy year for AI
advancement & adoption.”
The most well known AI tools are
those from OpenAI, such as image generator DALL-E 2 and text generator ChatGPT,
but the tech is advancing so quickly that by the time a certain industry has
grasped the implications of the latest development, another has emerged to
leapfrog in sophistication.
“Now it’s beginning to head toward video, and then
it’ll go 3D,” Mark Curtis, head of innovation at Accenture’s Interactive
division, tells Patrick Kulp at Adweek. “We’ve had to continuously rewrite this trend
over the last month and a half because new stuff was coming up. And I worry
that everything we’re going to say is going to be irrelevant by February.”
While imagery and text were the big
leaps forward in 2022, there are many other areas where machine learning
techniques could be on the brink of industry-transforming breakthroughs
including: music composition, video animation, writing code, and translation.
“It’s hard to guess which dominoes will fall first,
but by the end of this year, I don’t think artists will be alone in grappling
with their industry’s sudden automation,” says Vox’s
Kelsey Piper.
He predicts we’ll soon have image
models that can depict multiple characters or objects and consistently do more
complicated modeling of object interactions (a weakness of current systems).
“I doubt they’ll be perfect, but I
suspect most complaints about the limits of current systems will no longer
apply.”
Piper also suggests better text
generators — ones that provide better answers to nearly every question you ask
them. That may already be happening. Microsoft is reportedly planning to
integrate ChatGPT into its Bing search engine.
“Instead of providing links in
response to search queries, a language model-powered search engine could simply
answer questions.”
Marketers also say 2023 will be the
year that brands and agencies get serious about how synthetic content can be
deployed to serve bottom lines and augment human creativity.
“The things that agencies
should be doing is beyond experimenting with this; they should be calculating
now what it means for their business,” Curtis tells Kulp.
Generative AI, he added, “is a tool
humans will use to kickstart creative thinking or to create the base level of
something, which they then adapt continuously, or to move more quickly. …It is
not an answer to everything, but it does radically shift the economics of a lot
of what we do in creativity.”
Agency BBDO has experimented and
agrees that the ad industry should be thinking more about the various ways it
could revolutionize how creatives do their jobs.
“In my mind, it doesn’t appear that many of the
people commenting on this have even used the tool,” Zach Kula, group strategy
director, tells Adweek. “If they did, it would be obvious it’s not
even close to replacing creative thinking. In fact, I’d say it exposes how
valuable true creative thinking actually is. It puts the difference between
original creative thought and eloquently constructed database information in
plain sight.”
Experts say it’s likely that
technology like voice cloning, synthetic imagery and generated copy could align
in the next year to allow marketers to create full realistic-seeming videos out
of whole cloth with AI.
According to Kulp, those capabilities
could make it easier for marketers to make targeted, personalized video ads
aimed at different segments at scale.
In addition to possible upsides,
generative AI also has a host of risks that any marketer needs to be aware of,
including the potential for accidental copyright infringement or plagiarism.
Brands are already preparing defenses against fake content such as
auto-generated user reviews or defamatory content generated at scale.
Within five years, 80% of enterprise
marketers will establish a “dedicated content authenticity function” to root
out AI-generated misinformation, according to industry analyst Gartner. The
consultancy also projects that 70% of enterprise CMOs will list “accountability
in ethical AI” among their top concerns as more regulations and risks develop.
In fact, 2023 will be marked by a
tightening of regulations around AI. In the US, Microsoft (an investor in
OpenAI), GitHub and OpenAI are being sued in a class-action lawsuit that
accuses them of violating copyright law by letting Copilot, GitHub’s code
writing service, regurgitate sections of licensed code without providing
credit.
In Europe, the EU’s proposed AI Act could limit the
type of research that produces AI tools like GPT-3, experts have warned.
According to a TechCrunch article by Kyle Wiggers, so could more local efforts, like New York City’s
AI hiring statute, which requires that AI and algorithm-based tech for
recruiting, hiring or promotion be audited for bias before being used.
“Next year will only bring the threat
of regulation, though — expect much more quibbling over rules and court cases
before anyone gets fined or charged,” Wiggers says. “But companies may still
jockey for position in the most advantageous categories of upcoming laws, like
the AI Act’s risk categories.”
Brockman’s tweet is actually alarming
given the rapid advance of the technology and the failure of rules and ethical
considerations to keep pace with it.
“I think a slow, sleepy year on the
AI front would be good news for humanity,” Wiggers says. “We’d have some time
to adapt to the challenges AI poses, study the models we have, and learn about
how they work and how they break.
“And… we might have time for a more
serious conversation about why AI matters so much and how we — a human
civilization with a shared stake in this issue — can make it go well.”
Synthetic content generators are going to seem
trivial in comparison to the broader sweep of AI which is to effectively mimic
human intelligence. A human-level AI would be what Max Roser, founder and director of Our World in
Data, describes as a machine, or a network of machines, capable of carrying out
the same range of tasks that humans can.
Not so long ago the stuff of science
fiction, the date for such a development actually happening has been brought a
lot closer.
According to a number of experts and
surveys, including by the Metaculus community and research by Ajeya Cotra, who
works for the nonprofit Open Philanthropy, there is large agreement that the
timelines for achieving human-level AI are shorter than a century, and many
have timelines that are substantially shorter than that.
In Roser’s article, the majority of those who study this question believe that there is a 50% chance that transformative AI systems will be developed within the next 50 years. In this case, Roser says, it would plausibly be the biggest transformation in the lifetimes of our children, or even in our own lifetimes.
No comments:
Post a Comment