eMusings
Your eyes and ears on the worlds of art, culture, technology, philosophy - whatever stimulates the mind and excites the imagination. We remind you that 20 years of
back issues of eMusings can be found on our archives page.
By now you are familiar with the AI drumbeat surrounding us. Here are a few of the more informative sites:
The inborn biases of some LLMs (Large Language Models) are revealed in this
study. Chat AI, for example, contains hidden prejudices based on dialect, and these cannot be eliminated with standard
procedures. In testing, it was found that some AI systems will recommend the death penalty for fictional defendants who use
African American English. The systems will also offer less prestigious jobs to these same applicants. The potential for harm
was described as "massive". Using human feedback to correct these biases had no effect and was described by a researcher as
as attempt to "simply paper over the rot".
A new 4-legged
robot has been able to overcome previously unlearned obstacles in real-life situations. This activity moves far beyond
the highly-programmed robots we have seen from Boston Dynamics, a leader in robotic development. The new robot was designed
using simulation, rather than physical tasks, and involves neural networks instead of standard software. Trial and error were
fundamental to the training, as opposed to human demonstrations.
Cognition Labs has presented Devin, described as
the world's "first AI software engineer". Devin, we are told, can build and use entire projects on its own. It can also find
and repair bugs, and it trains and uses its own AI systems to solve difficulties. Essentially, Devin acts like a complete
software team: tell it what you want to accomplish and it will come up with a plan, create little bots to execute the plan,
and finally test and debug until the system meets your demands. More problematic is Devin's ability to make and train
its own robot slave from start to finish within a couple of hours.
You have probably heard of Sora,
the new AI generator that produces videos. Still not available to the general public, this link will show you some videos created by
the Sora team.
Midjourney has created a new feature for its image-generation AI. Now you can re-use the same character in multiple images.
Substantial research is being devoted to robots that can heal
themselves. Already several self-healing electronic components have been developed; the challenge is to integrate them into machines.
Some healing requires a specific signal, called non-autonomous healing. Others seem not to require a trigger. Sometimes as well it
is preferable for a robot not to heal immediately - ie the work it is doing must be completed first, or the humidity around it is not
suitable for healing. Also, as robots become more complex, multiple layers are involved, complicating the healing process.
A new humanoid called H1 created by
China's Unitree Robotics can climb stairs and has broken a speed record by moving at 7.4 mph. The robot is 71" tall and can carry
weights up to 66 lbs. It also dances and can jump as high as a man next to it. Incidentally, if you want one of these robots for
yourself, it won't be available for somewhere between 3 - 10 years and then might sell for $90,000.
Anthropic claims that its new family of Claude 3 models can outperform both Google's Gemini and Open AI's ChatGPT. Anthropic's
previous iterations of Claude had difficulty answering some questions. but they are now saying that the new Claude family
"exhibits near-human levels of comprehension… [for] complex tasks". The new Claudes, they tell us, can solve math problems, write computer code,
and reason well. Of course there is still the problem of "hallucinations" - ie responses with wrong information.
Creating video
games is the next challenge facing AI algorithms. Google Deep Mind tells us that Genie, its new AI, was trained on 30,000 hours of select
video footage amd now can create games on its own. Genie can apparently turn a sketch, a photo, or an image into an interactive game. The AI
video games are currrently slow - one frame per second compared to modern video games that hit 60 or 120 frames per second. Like other
AI models, it does tend to create strange visual "hallucinations", an elegant euphemism for false results.
Now on to other April treats:
A new survey of Black
Portraiture takes place at the National Portrait Gallery in London. Titled, "The Time is Always Now: Artists Reframe the Black figure",
the exhibit features 22 artists who contemplate "a history of being overlooked, misrepresented, or depicted without agency". There are
excellent examples here of always being on the outside, under the "white gaze". Reference is also made to the 2022 outstanding
show, "In the Black Fantastic".
"Unit" is derived from AI machine learning combined with
other digital imaging tools to portray a single dancer moving through time and space. A series of silhouettes is fed into a
neural network which alters the images into rectangular structures. Called a "shape language", this code was developed
to teach the algorithm to transform round organic shapes into rectilinear abstractions. Note: this is the exact opposite
of what early digital imagists like myself struggled with in the nascent decades of digital painting and sculpture. At that
pioneering time, computers and engineers were caught in rectangular image-making, where edges and corners prevailed over rounded
and organic shapes. For me personally it was a profound struggle to develop a language of understanding so that basically
mechanical engineers would agree to output my vision rather than their own. I was fortunate to find open-minded fabricators who
were willing to abaondon the box and experiment with me. I salute those men, since no women were activly involved in digital fabrication.
.
Opening in time for the Venice Biennale, Italy's most extensive survey of Willem de
Kooning follows the artist's 2 visits to Italy in 1959 and 1969. It is a treat to see the glorious color and flow of these
paintings. The curators write, "Willem de Kooning collected from the cacophony of visual excitement, light and movement in daily life to
create his own lexicon."
"Speak to the Eye" presents a
retrospective of early digital works by Vera Molnar, who died 2 months ago. Shown at Paris' Centre Pompidou. Molnar
called her works "Machine Imaginaire" as she used simple algorithms to experiment with the placement of shapes and lines.
A new biography of Keith
Haring asks whether the artist sold his soul to capitalism and commercialism. It is true that innumerable products and
gadgets can be bought with Haring's works, from T shirts to cups to magnets to skateboards. Does that mean he abandoned
pure art, whatever that is? Comments range from the ultranegative ("what precipitated Haring’s eclipse was not so much the suspicion that he had prostituted his art, but that he had
nothing to prostitute", Tad Friend) to the praise (Haring "(brought) back the ideals of conceptual art by spinning the
stuff of the everyday into works galleries could show and sell", Dennis Hopper).
Anne Samat creates totemic works from woven pieces
and symbols. She uses the Southeast Asian tradition of Pua Kumbu
weaving along with items found in 99 cent stores to
embody history within the sculptural tradition. The works are highly detailed, both personal and generic, and strongly evocative.
Edward Burtynsky celebrates the glory of nature in his abstractions,
commenting, "We come from nature. There is an importance to [having] a certain reverence for what nature is because we are
connected to it... If we destroy nature, we destroy ourselves." His large-scale photographs reveal the good, the bad, and
the ugly in our environment. His Anthropocene Project combines science, film, photography, AI and AR to examine our
unique impact upon the planet and what it means for the future of the Earth.
"A Cacophony of
Rocks" is the title of sculptor Heidi Lau's first solo exhibition at Sikkema Jenkins in New York. Lau was inspired by the
classical stories of Shanhaijing, Chinese mythology, and Taoist traditions. Ceramic vessels are shown with animals and
flowers, glass and bronze, embodying the idea that death is somehow "permeable", and the past and the present can intermingle. Her
use of the antilinear produces a sense of another world, beyond the one we experience every day.
c. Corinne Whitaker 2024
want to know more about the art?
front page , new
paintings, new
blobs, new
sculpture, painting
archives, blob
archives, sculpture
archives,
photography archives, Archiblob archives, image
of the month, blob
of the month,
art headlines,
technology news, electronic
quill, electronic
quill archives, art smart
quiz, world art
news,
eMusings,
eMusings archive, readers
feast, whitaker
on the web,
nations one,
meet the giraffe,
studio map, just
desserts, Site
of the Month, young
at art,
about the artist?
copyright 2024 Corinne Whitaker