Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Deep Seek has been at the center of global artificial
intelligence news this week, continuing to capture attention for its
striking technical advances and rapid international growth. In early twenty
twenty five, deep Seek, a Chinese start up, surprised both
industry leaders and government officials with its r IE language model,
which not only matched the reasoning capacity of top American systems,
(00:22):
but did so with unprecedented cost efficiency. The Washington Times
highlights how deep Seek's work was seen as a wake
up call for U S tech companies and security agencies,
with its r IE model demonstrating reasoning breakthroughs that even
OpenAI believed would remain an American domain for several more months.
The R one launch stunned the market by achieving wide
(00:43):
adoption almost overnight. Within one week, Deepseek's mobile app climbed
the top of the U S App Store, surpassing open
AI's Chat GPT for the first time since twenty twenty two.
As reported by American Bazaar, This caused historic disruption in
texter In Nvidia, a crucial supplier of AI hardware, lost
(01:04):
nearly nine percent in a single day, wiping out three
hundred billion dollars in market value before quickly rebounding as
analysts recognized deep Seek's innovations were driving new opportunities in
the hardware market rather than threatening it entirely. The R
one model's most notable feature remains its efficiency. Deep Seek
(01:26):
claims it cost under six million dollars to train, compared
to the roughly six hundred million dollars to one billion
dollars figures required for open AI's frontier models. Security research
from Strider and the Special Competitive Studies Project points to
another area of global interest. Deep Seek enabled assistant systems
(01:48):
are now supported by China's Heinan subse intelligent Computing Center.
This underwater facility, operating about thirty yards below sea level,
runs AI models that handle up to SI seven thousand
conversations per second. The center is a joint project involving
companies under US sanctions and is linked to organizations with
(02:09):
ties to Chinese military and state security. Intelligence experts believe
this infrastructure reflects not only China's commercial ambitions, but its
broader military and national security strategy, Underscoring the dual use
nature of advanced AI technologies. American researchers are examining whether
(02:29):
Chinese technological strategies differ substantially from the dominant approaches in
the United States. While U S labs have focused heavily
on large scale statistical language models, many in China have
pursued more biologically inspired algorithms, some modeled after human cognitive
or neural processes. At an AI conference in Washington last month,
(02:53):
Open AI's Katrina Mulligan publicly acknowledged that deep seeks leap
forward had forced a rethink king of assumptions about the
global AI race, noting that her team's previous confidence in
their lead had been shaken by deep Seek's achievements. Apple's
research division has entered the debate as well, publishing a
new paper challenging the widespread idea that current large language models,
(03:18):
including deep Seeks, are on a straight trajectory to artificial
general intelligence, pointing to key conceptual and technical bottlenecks. In
a sign of how quickly the landscape is changing, Deep
Seek now faces competition from other Chinese AI outfits. Zai,
for example, has released its own open source GLM four
(03:40):
point five model and claims even greater price performance. CNBC
reported that Zai's system costs just eleven cents per million
input tokens compared to deep Seek's fourteen cents, with output
costs also significantly lower. Z has recently ran raised one
dollar and five cents and is planning a high profile
(04:04):
IPO inside China, showing the scale and velocity of capital
flowing into the sector. Amid these breakthroughs, national policy responses
are accelerating. The Trump administration has rolled out an AI
Action Plan seeking to ensure US leadership in AI, while
biparties in policy panels advocate for a Manhattan Project style
(04:27):
initiative to advance American progress in artificial general intelligence. The
continuing advances out of China and deep Seek's influence on
global markets, technology strategy, and even national policy highlight how
the AI landscape is changing by the day. According to
(04:47):
crypto Neews, deep Seek's predictive analytics tools have even been
put to the test forecasting cryptocurrency prices, showing how its
technology is being rapidly commercialized across new domains. Thanks for
tuning in to this week's update on deep Seek and
the global AI race. Come back next week for more.
(05:09):
This has been a quiet please production. For more check
out Quiet Please dot a I