Earlier this month, DeepMind introduced a brand new “generalist” AI mannequin known as Gato. The mannequin can play the online game Atari, caption pictures, chat, and stack blocks with an actual robotic arm, the Alphabet-owned AI lab introduced. All in all, Gato can do tons of of various duties.
However whereas Gato is undeniably fascinating, within the week since its launch some researchers have gotten a bit carried away.
One among DeepMind’s high researchers and a coauthor of the Gato paper, Nando de Freitas, couldn’t include his pleasure. “The sport is over!” he tweeted, suggesting that there’s now a transparent path from Gato to synthetic normal intelligence, or ‘AGI’, a obscure idea of human or superhuman-level AI. The best way to construct AGI, he claimed, is usually a query of scale: making fashions equivalent to Gato greater and higher.
Unsurprisingly, de Freitas’s announcement triggered breathless press protection that Deepmind is “on the verge” of human-level synthetic intelligence. This isn’t the primary time hype has outstripped actuality. Different thrilling new AI fashions, equivalent to OpenAI’s textual content generator GPT-3 and picture generator DALL-E, have generated comparable grand claims.
For a lot of within the subject, this sort of feverish discourse overshadows different necessary analysis areas in AI. Learn the complete story.
I’ve combed the web to seek out you in the present day’s most enjoyable/necessary/scary/fascinating tales about know-how.
1 Volunteers are translating Chinese language social media posts into English
Despite the fact that the posts have handed China’s web censorship regime, Beijing is sad. (The Atlantic $)
+ WeChat desires folks to make use of its video platform. So that they did, for digital protests. (TR)
2 Ukraine’s startup group is resuming enterprise as standard
Many employees are juggling their day jobs with after-hours struggle effort volunteering. (WP $)
+ Russian-speaking tech bosses residing within the US are reducing ties with pro-war employees. (NYT $)
+ YouTube has taken down greater than 9,000 channels linked to the struggle. (The Guardian)
3 The Buffalo taking pictures highlighted the failings of tech’s anti-terrorism accord
Critics say platforms haven’t finished sufficient to sort out the basis causes of extremism. (WSJ $)
+ America has skilled greater than 3,500 mass shootings since Sandy Hook. (WP $)
4 Crypto seems to have an insider buying and selling drawback
Identical to the banking system its supporters rail in opposition to. (WSJ $)
+ Christine Lagarde thinks crypto is price “nothing.” (Bloomberg $)
+ Crypto is weathering a bitter storm. Some nonetheless maintain on for pricey life. (TR)
+ The crypto trade has misplaced round $1.5 trillion since November. (The Atlantic $)
+ Stablecoin Tether has paid out $10 billion in withdrawals for the reason that crash began. (The Guardian)
5 The nuclear fusion trade is in turmoil
It isn’t even up and operating but, however gas provides are already operating low. (Wired $)
+ A gap within the floor could possibly be the way forward for fusion energy. (TR)
+ The US midwest could possibly be dealing with energy grid failure this summer time. (Motherboard)
6 Massive Tech isn’t nervous concerning the financial downturn
Even when it drops a few of its market valuation alongside the best way. (NYT $)
+ However lawmakers are decided to rein them in with antitrust laws. (Recode)
+ Their carbon emissions are spiraling uncontrolled, too. (New Yorker $)
7 The US army desires to construct a flying ship
The Liberty Lifer X-plane can be unbiased of mounted airfields and ports. (IEEE Spectrum)
8 We have to change how we recycle plastic
The excellent news is that the know-how to overtake it exists—it simply wants refining. (Wired $)
+ A French firm is utilizing enzymes to recycle one of the vital widespread single-use plastics. (TR)
9 Why it’s best to deal with utilizing your telephone like consuming wine
Placing that delicate steadiness from stopping the constructive tipping into unfavorable. (The Guardian $)
Quote of the day
“I like the moment gratification of creating the web higher.”
—Jason Moore, who’s credited with creating greater than 50,000 Wikipedia pages, tells CNN about his motivations for taking up the unpaid work.