Earth’s Core Has Stopped and May Be Reversing Direction, Study Says
The surprising finding might solve longstanding mysteries about climate and geological phenomena.
https://www.vice.com/en/article/xgyje7/ ... study-says
Earth’s inner core has recently stopped spinning, and may now be reversing the direction of its rotation, according to a surprising new study that probed the deepest reaches of our planet with seismic waves from earthquakes.
The mind-boggling results suggest that Earth’s center pauses and reverses direction on a periodic cycle lasting about 60 to 70 years, a discovery that might solve longstanding mysteries about climate and geological phenomena that occur on a similar timeframe, and that affect life on our planet.
Of course, it must be noted this is more or less the plot of the 2003 disaster film The Core, but there’s no need to worry about averting an impending apocalypse by nuking the center of Earth. While the core’s rotation influences Earth’s surface environment, scientists think this periodic spin switch is a normal part of its behavior that does not pose risks for life on our planet.
Earth’s inner core is a solid metal ball that is 75 percent the size of the Moon. It can spin at different speeds and directions compared to our planet because it is nestled within a liquid outer core, but scientists are not sure exactly how fast it spins or whether its speed varies over time.
Located some 3,000 miles beneath our feet, the core experiences intense heat on par with the surface of the Sun. Because it is so remote and difficult to study, the inner core remains one of the least understood environments on our planet, though it’s clear that it plays a role in many processes that make our world habitable to life, such as the generation of Earth’s protective magnetic field, which blocks harmful radiation from reaching the surface.
Now, Yi Yang and Xiaodong Song, a pair of researchers at Peking University’s SinoProbe Lab at School of Earth and Space Sciences, have captured “surprising observations that indicate the inner core has nearly ceased its rotation in the recent decade and may be experiencing a turning-back in a multidecadal oscillation, with another turning point in the early 1970s,” according to a study published on Monday in Nature Geoscience.
“There are two major forces acting on the inner core,” Yang and Song said in an email to Motherboard. “One is the electromagnetic force. The Earth’s magnetic field is generated by fluid motion in the outer core. The magnetic field acting on the metallic inner core is expected to drive the inner core to rotate by electromagnetic coupling. The other is gravity force. The mantle and inner core are both highly heterogeneous, so the gravity between their structures tends to drag the inner core to the position of gravitational equilibrium, so called gravitational coupling.”
“If the two forces are not balanced out, the inner core will accelerate or decelerate,” they added. “Both the magnetic field and the Earth’s rotation have a strong periodicity of 60-70 years. We believe that the proposed 70-year oscillation of the inner core is driven by the electromagnetic and gravitational forces.”
Song has spent decades trying to unravel the mysteries of the inner core by studying seismic waves that pass through this distant region. He was part of the team that first reported evidence of the inner core’s rotation in 1996 by measuring slight time (or “temporal”) changes in these waves, which are generated by earthquakes.
However, the origin of the temporal changes has been a matter of debate within the geoscience community ever since, as some scientists think the wave patterns arise from phenomena at the boundary between the outer and inner core.
“Some researchers are still arguing that the temporal changes do not come from the inner-core rotation, but from localized deformation at the inner core boundary,” Yang and Song said. With their new study, the pair “tried to gather more data over a longer duration to test different models.”
To that end, the team studied seismic waves that passed through the inner core made by earthquakes that occurred since the 1960s. In particular, they looked for “doublet” events, which are “repeating earthquakes with nearly identical waveforms at common receivers,” according to the study. By analyzing the slight temporal changes between these doublets, Yang and Song were able to probe the rotation of the inner core.
As it turned out, the temporal changes reached a minimum around 2009, suggesting that the inner core had paused rotation around this time, creating seismic observations that seem more static. The team was even more astonished when they identified a similar turning point in the early 1970s, hinting that the core stops and reverses rotation on a periodic cycle.
“Our results further support the inner-core rotation, and more interestingly, reveals the multidecadal pattern of the rotation,” Yang and Song said to Motherboard.
The results offer an unprecedented look at the searing pit of our planet, a region that continues to evade clear explanation, and it also has big implications for understanding the familiar world we inhabit on Earth’s surface.
For instance, the team notes that the same multidecade cycle has also been observed in Earth’s climate system, as global mean temperatures and sea level rises appear to oscillate every 60 to 70 years. The length of Earth’s day, which shifts slightly over time, also seems synced to the proposed cycle. For this reason, the new findings “may imply dynamic interactions between the deepest and shallowest layers of the solid Earth system,” according to the new study.
“We pointed the existence of similar periodicity of different observations, forming a
resonating system,” Yang and Song told Motherboard. “The linkage, however, is less clear at the moment. The gravitational coupling between the inner core and the mantle may cause deformation at the Earth’s surface, which would affect the sea level. The changes of the sea level and the Earth’s rotation may affect the global atmosphere circulation and temperature. The resonance of different systems may also amplify the mutual interactions.”
It’s tantalizing to imagine that our most mundane experiences—such as the length of our days, and the climatic patterns that guide our local weather—might be sculpted by the rotational cycles of a weird metal ball at the center of our world. Untangling these nuances will require new models and continued observations of Earth’s enigmatic central realm.
The next steps are “to build quantitative models of the physical mechanisms on the multi-decadal oscillation system” and “to monitor how the rotation changes in the future,” Yang and Song said.
“We’d expect it to rotate westwards relative to the surface of the Earth in the coming years and decades,” the pair concluded. “Seismic waves are still the best way and thus continuous operation of high-quality seismic networks is crucial in this regard.”
Nauka i tehnologija
Re: Nauka i tehnologija
Najgore je svađati se sa budalom. Prvo te spusti na svoj nivo, a onda te dotuče iskustvom...
Re: Nauka i tehnologija
Sent from my SM-S906B using Tapatalk
Re: Nauka i tehnologija
In a First, Caltech's Space Solar Power Demonstrator Wirelessly Transmits Power in Space
https://www.caltech.edu/about/news/in-a ... r-in-space
A space solar power prototype that was launched into orbit in January is operational and has demonstrated its ability to wirelessly transmit power in space and to beam detectable power to Earth for the first time.
Wireless power transfer was demonstrated on March 3 by MAPLE, one of three key technologies being tested by the Space Solar Power Demonstrator (SSPD-1), the first space-borne prototype from Caltech's Space Solar Power Project (SSPP). SSPP aims to harvest solar power in space and transmit it to the Earth's surface.
MAPLE, short for Microwave Array for Power-transfer Low-orbit Experiment and one of the three key experiments within SSPD-1, consists of an array of flexible lightweight microwave power transmitters driven by custom electronic chips that were built using low-cost silicon technologies. It uses the array of transmitters to beam the energy to desired locations. For SSPP to be feasible, energy transmission arrays will need to be lightweight to minimize the amount of fuel needed to send them to space, flexible so they can fold up into a package that can be transported in a rocket, and a low-cost technology overall.
MAPLE was developed by a Caltech team led by Ali Hajimiri, Bren Professor of Electrical Engineering and Medical Engineering and co-director of SSPP.
"Through the experiments we have run so far, we received confirmation that MAPLE can transmit power successfully to receivers in space," Hajimiri says. "We have also been able to program the array to direct its energy toward Earth, which we detected here at Caltech. We had, of course, tested it on Earth, but now we know that it can survive the trip to space and operate there."
Using constructive and destructive interference between individual transmitters, a bank of power transmitters is able to shift the focus and direction of the energy it beams out—without any moving parts. The transmitter array uses precise timing-control elements to dynamically focus the power selectively on the desired location using the coherent addition of electromagnetic waves. This enables the majority of the energy to be transmitted to the desired location and nowhere else.
MAPLE features two separate receiver arrays located about a foot away from the transmitter to receive the energy, convert it to direct current (DC) electricity, and use it to light up a pair of LEDs to demonstrate the full sequence of wireless energy transmission at a distance in space. MAPLE tested this in space by lighting up each LED individually and shifting back and forth between them. The experiment is not sealed, so it is subject to the harsh environment of space, including the wide temperature swings and solar radiation that will be faced one day by large-scale SSPP units.
"To the best of our knowledge, no one has ever demonstrated wireless energy transfer in space even with expensive rigid structures. We are doing it with flexible lightweight structures and with our own integrated circuits. This is a first," says Hajimiri.
MAPLE also includes a small window through which the array can beam the energy. This transmitted energy was detected by a receiver on the roof of the Gordon and Betty Moore Laboratory of Engineering on Caltech's campus in Pasadena on May 22. The received signal appeared at the expected time and frequency, and had the right frequency shift as predicted based on its travel from orbit.
Beyond a demonstration that the power transmitters could survive the launch (which took place on January 3) and space flight, and still function, the experiment has provided useful feedback to SSPP engineers. The power transmission antennas are clustered in groups of 16, each group driven by one entirely custom flexible integrated circuit chip, and Hajimiri's team now is assessing the performance of individual elements within the system by evaluating the interference patterns of smaller groups and measuring difference between various combinations. The painstaking process—which can take up to six months to fully complete—will allow the team to sort out irregularities and trace them back to individual units, providing insight for the next generation of the system.
Space solar power provides a way to tap into the practically unlimited supply of solar energy in outer space, where the energy is constantly available without being subjected to the cycles of day and night, seasons, and cloud cover—potentially yielding eight times more power than solar panels at any location on Earth's surface. When fully realized, SSPP will deploy a constellation of modular spacecraft that collect sunlight, transform it into electricity, then convert it to microwaves that will be transmitted wirelessly over long distances to wherever it is needed—including locations that currently have no access to reliable power.
"The flexible power transmission arrays are essential to the current design of Caltech's vision for a constellation of sail-like solar panels that unfurl once they reach orbit," says Sergio Pellegrino, Joyce and Kent Kresa Professor of Aerospace and Civil Engineering and co-director of SSPP.
"In the same way that the internet democratized access to information, we hope that wireless energy transfer democratizes access to energy," Hajimiri says. "No energy transmission infrastructure will be needed on the ground to receive this power. That means we can send energy to remote regions and areas devastated by war or natural disaster."
SSPP got its start in 2011 after philanthropist Donald Bren, chairman of Irvine Company and a lifetime member of the Caltech Board of Trustees, first learned about the potential for space-based solar energy manufacturing as a young man in an article in the magazine Popular Science. Intrigued by the potential for space solar power, in 2011, Bren approached Caltech's then-president Jean-Lou Chameau to discuss the creation of a space-based solar power research project. In the years to follow, Bren and his wife, Brigitte Bren, also a Caltech trustee, agreed to make the donation to fund the project. The first of the donations to Caltech (which will eventually exceed $100 million in support for the project and endowed professorships) was made through the Donald Bren Foundation.
"The hard work and dedication of the brilliant scientists at Caltech have advanced our dream of providing the world with abundant, reliable and affordable power for the benefit of all humankind," Bren says.
"The transition to renewable energy, critical for the world's future, is limited today by energy storage and transmission challenges. Beaming solar power from space is an elegant solution that has moved one step closer to realization due to the generosity and foresight of the Brens," says Caltech President Thomas F. Rosenbaum. "Donald Bren has presented a formidable technical challenge that promises a remarkable payoff for humanity: a world powered by uninterruptible renewable energy."
In addition to the support received from the Brens, Northrop Grumman Corporation also provided Caltech $12.5 million over three years through a sponsored research agreement between 2014 and 2017 that supported for the development of technology and advancement of science for the project.
"Demonstration of wireless power transfer in space using lightweight structures is an important step toward space solar power and broad access to it globally," says Harry Atwater, Otis Booth Leadership Chair of Division of Engineering and Applied Science; Howard Hughes Professor of Applied Physics and Materials Science; Director of the Liquid Sunlight Alliance; and one of the principal investigators of the project. "Solar panels already are used in space to power the International Space Station, for example, but to launch and deploy large enough arrays to provide power to Earth, SSPP has to design and create solar power energy transfer systems that are ultra-lightweight, cheap, and flexible."
Individual SSPP units will fold up into packages about 1 cubic meter in volume and then unfurl into flat squares about 50 meters per side, with solar cells on one side facing toward the sun and wireless power transmitters on the other side facing toward Earth.
A Momentus Vigoride spacecraft launched aboard a SpaceX rocket on the Transporter-6 mission carried 50-kilogram SSPD to space. Momentus is providing ongoing hosted payload support to Caltech, including providing data, communication, commanding and telemetry, and resources for optimal picture taking and solar cell lighting. The entire set of three prototypes within the SSPD was envisioned, designed, built, and tested by a team of about 35 individuals—faculty, postdocs, graduate students, and undergrads—in labs at Caltech.
SSPD has two main experiments besides MAPLE: DOLCE (Deployable on-Orbit ultraLight Composite Experiment), a structure measuring 6 feet by 6 feet that demonstrates the architecture, packaging scheme, and deployment mechanisms of the modular spacecraft; and ALBA, a collection of 32 different types of photovoltaic cells to enable an assessment of the types of cells that are the most effective in the punishing environment of space. The ALBA tests of solar cells are ongoing, and the SSPP has not yet attempted to deploy DOLCE as of press time. Results from those experiments are expected in the coming months.
https://www.caltech.edu/about/news/in-a ... r-in-space
A space solar power prototype that was launched into orbit in January is operational and has demonstrated its ability to wirelessly transmit power in space and to beam detectable power to Earth for the first time.
Wireless power transfer was demonstrated on March 3 by MAPLE, one of three key technologies being tested by the Space Solar Power Demonstrator (SSPD-1), the first space-borne prototype from Caltech's Space Solar Power Project (SSPP). SSPP aims to harvest solar power in space and transmit it to the Earth's surface.
MAPLE, short for Microwave Array for Power-transfer Low-orbit Experiment and one of the three key experiments within SSPD-1, consists of an array of flexible lightweight microwave power transmitters driven by custom electronic chips that were built using low-cost silicon technologies. It uses the array of transmitters to beam the energy to desired locations. For SSPP to be feasible, energy transmission arrays will need to be lightweight to minimize the amount of fuel needed to send them to space, flexible so they can fold up into a package that can be transported in a rocket, and a low-cost technology overall.
MAPLE was developed by a Caltech team led by Ali Hajimiri, Bren Professor of Electrical Engineering and Medical Engineering and co-director of SSPP.
"Through the experiments we have run so far, we received confirmation that MAPLE can transmit power successfully to receivers in space," Hajimiri says. "We have also been able to program the array to direct its energy toward Earth, which we detected here at Caltech. We had, of course, tested it on Earth, but now we know that it can survive the trip to space and operate there."
Using constructive and destructive interference between individual transmitters, a bank of power transmitters is able to shift the focus and direction of the energy it beams out—without any moving parts. The transmitter array uses precise timing-control elements to dynamically focus the power selectively on the desired location using the coherent addition of electromagnetic waves. This enables the majority of the energy to be transmitted to the desired location and nowhere else.
MAPLE features two separate receiver arrays located about a foot away from the transmitter to receive the energy, convert it to direct current (DC) electricity, and use it to light up a pair of LEDs to demonstrate the full sequence of wireless energy transmission at a distance in space. MAPLE tested this in space by lighting up each LED individually and shifting back and forth between them. The experiment is not sealed, so it is subject to the harsh environment of space, including the wide temperature swings and solar radiation that will be faced one day by large-scale SSPP units.
"To the best of our knowledge, no one has ever demonstrated wireless energy transfer in space even with expensive rigid structures. We are doing it with flexible lightweight structures and with our own integrated circuits. This is a first," says Hajimiri.
MAPLE also includes a small window through which the array can beam the energy. This transmitted energy was detected by a receiver on the roof of the Gordon and Betty Moore Laboratory of Engineering on Caltech's campus in Pasadena on May 22. The received signal appeared at the expected time and frequency, and had the right frequency shift as predicted based on its travel from orbit.
Beyond a demonstration that the power transmitters could survive the launch (which took place on January 3) and space flight, and still function, the experiment has provided useful feedback to SSPP engineers. The power transmission antennas are clustered in groups of 16, each group driven by one entirely custom flexible integrated circuit chip, and Hajimiri's team now is assessing the performance of individual elements within the system by evaluating the interference patterns of smaller groups and measuring difference between various combinations. The painstaking process—which can take up to six months to fully complete—will allow the team to sort out irregularities and trace them back to individual units, providing insight for the next generation of the system.
Space solar power provides a way to tap into the practically unlimited supply of solar energy in outer space, where the energy is constantly available without being subjected to the cycles of day and night, seasons, and cloud cover—potentially yielding eight times more power than solar panels at any location on Earth's surface. When fully realized, SSPP will deploy a constellation of modular spacecraft that collect sunlight, transform it into electricity, then convert it to microwaves that will be transmitted wirelessly over long distances to wherever it is needed—including locations that currently have no access to reliable power.
"The flexible power transmission arrays are essential to the current design of Caltech's vision for a constellation of sail-like solar panels that unfurl once they reach orbit," says Sergio Pellegrino, Joyce and Kent Kresa Professor of Aerospace and Civil Engineering and co-director of SSPP.
"In the same way that the internet democratized access to information, we hope that wireless energy transfer democratizes access to energy," Hajimiri says. "No energy transmission infrastructure will be needed on the ground to receive this power. That means we can send energy to remote regions and areas devastated by war or natural disaster."
SSPP got its start in 2011 after philanthropist Donald Bren, chairman of Irvine Company and a lifetime member of the Caltech Board of Trustees, first learned about the potential for space-based solar energy manufacturing as a young man in an article in the magazine Popular Science. Intrigued by the potential for space solar power, in 2011, Bren approached Caltech's then-president Jean-Lou Chameau to discuss the creation of a space-based solar power research project. In the years to follow, Bren and his wife, Brigitte Bren, also a Caltech trustee, agreed to make the donation to fund the project. The first of the donations to Caltech (which will eventually exceed $100 million in support for the project and endowed professorships) was made through the Donald Bren Foundation.
"The hard work and dedication of the brilliant scientists at Caltech have advanced our dream of providing the world with abundant, reliable and affordable power for the benefit of all humankind," Bren says.
"The transition to renewable energy, critical for the world's future, is limited today by energy storage and transmission challenges. Beaming solar power from space is an elegant solution that has moved one step closer to realization due to the generosity and foresight of the Brens," says Caltech President Thomas F. Rosenbaum. "Donald Bren has presented a formidable technical challenge that promises a remarkable payoff for humanity: a world powered by uninterruptible renewable energy."
In addition to the support received from the Brens, Northrop Grumman Corporation also provided Caltech $12.5 million over three years through a sponsored research agreement between 2014 and 2017 that supported for the development of technology and advancement of science for the project.
"Demonstration of wireless power transfer in space using lightweight structures is an important step toward space solar power and broad access to it globally," says Harry Atwater, Otis Booth Leadership Chair of Division of Engineering and Applied Science; Howard Hughes Professor of Applied Physics and Materials Science; Director of the Liquid Sunlight Alliance; and one of the principal investigators of the project. "Solar panels already are used in space to power the International Space Station, for example, but to launch and deploy large enough arrays to provide power to Earth, SSPP has to design and create solar power energy transfer systems that are ultra-lightweight, cheap, and flexible."
Individual SSPP units will fold up into packages about 1 cubic meter in volume and then unfurl into flat squares about 50 meters per side, with solar cells on one side facing toward the sun and wireless power transmitters on the other side facing toward Earth.
A Momentus Vigoride spacecraft launched aboard a SpaceX rocket on the Transporter-6 mission carried 50-kilogram SSPD to space. Momentus is providing ongoing hosted payload support to Caltech, including providing data, communication, commanding and telemetry, and resources for optimal picture taking and solar cell lighting. The entire set of three prototypes within the SSPD was envisioned, designed, built, and tested by a team of about 35 individuals—faculty, postdocs, graduate students, and undergrads—in labs at Caltech.
SSPD has two main experiments besides MAPLE: DOLCE (Deployable on-Orbit ultraLight Composite Experiment), a structure measuring 6 feet by 6 feet that demonstrates the architecture, packaging scheme, and deployment mechanisms of the modular spacecraft; and ALBA, a collection of 32 different types of photovoltaic cells to enable an assessment of the types of cells that are the most effective in the punishing environment of space. The ALBA tests of solar cells are ongoing, and the SSPP has not yet attempted to deploy DOLCE as of press time. Results from those experiments are expected in the coming months.
Najgore je svađati se sa budalom. Prvo te spusti na svoj nivo, a onda te dotuče iskustvom...
Re: Nauka i tehnologija
Najgore je svađati se sa budalom. Prvo te spusti na svoj nivo, a onda te dotuče iskustvom...
-
shandu90
- Postovi: 12306
- Pridružio se: 22 Jan 2016, 23:13
- Garaža: Opel Astra J 1.4 Turbo
- Lokacija: Novi Sad
Re: Nauka i tehnologija
Nece da mi otvori retweet, trazi nalog. No, svaka im cast sta su uspeli da urade. Ostaje da se vidi sta ce biti u praksi.
Sent from my iPhone using Tapatalk
Sent from my iPhone using Tapatalk
Re: Nauka i tehnologija
Šefovi OpenAI-ja upozoreni na "otkriće koje će ugroziti čovječanstvo". Što je Q*?
https://www.index.hr/vijesti/clanak/sto ... i_ostalo_d
PRIJE nego što je Sam Altman dobio otkaz, OpenAI je navodno radio na naprednom sustavu koji je bio toliko moćan da je izazvao zabrinutost za sigurnost među zaposlenicima kompanije. Model umjetne inteligencije izazvao je takvu uzbunu kod nekih istraživača OpenAI-ja da su se obratili upravnom odboru prije Altmanovog otkaza upozoravajući da bi mogao ugroziti čovječanstvo, prenio je Reuters.
Model nazvan Q* - izgovara se kao "Q-Star" - bio je u stanju riješiti osnovne matematičke probleme s kojima se nikad prije nije susreo, prema stranici s tehnološkim vijestima Information, koja je dodala da je tempo razvoja sustava alarmirao neke sigurnosne stručnjake. "Uz pomoć golemih računalnih resursa, novi je model mogao riješiti određene matematičke probleme", rekao je izvor za Reuters. Sposobnost rješavanja matematičkih problema smatrala bi se značajnim razvojem umjetne inteligencije, piše The Guardian.
Izvještaji su uslijedili nakon višednevnih previranja u OpenAI-ju iz San Francisca, čiji je Upravni odbor otpustio Altmana prošlog petka, ali ga je vratio na posao u utorak navečer nakon što je gotovo svih 750 zaposlenika kompanije zaprijetilo odlaskom ako se ne vrati. Altman je također imao podršku najvećeg investitora OpenAI-ja - Microsofta.
Razvoj opće umjetne inteligencije?
Mnogi su stručnjaci zabrinuti da se kompanije kao što je OpenAI kreću prebrzo prema razvoju opće umjetne inteligencije (AGI), odnosno prema sustavu koji može obavljati širok raspon zadataka na ljudskoj razini ili iznad razine ljudske inteligencije - i koji bi u teoriji mogao izbjeći ljudsku kontrolu.
Andrew Rogoyski s Instituta za umjetnu inteligenciju usmjerenu na ljude na Sveučilištu Surrey, rekao je da bi postojanje velikog jezičnog modela (LLM) za rješavanje matematike bio napredak: "Intrinzična sposobnost LLM-a da rješava matematiku veliki je korak naprijed koji dopušta umjetnoj inteligenciji posve novi niz analitičkih sposobnosti".
Govoreći u četvrtak prošlog tjedna, dan prije iznenadnog otkaza, Altman je naznačio da je kompanija koja stoji iza ChatGPT-ja napravila još jedan iskorak. U govoru na summitu Azijsko-pacifičke gospodarske suradnje (APEC), rekao je sljedeće: "Četiri puta u povijesti OpenAI-ja, a posljednji put prije tek nekoliko tjedana, bio sam u prostoriji kada je na neki način skinut veo neznanja i granica otkrića pomaknuta naprijed. Profesionalna je čast sudjelovati u tome."
Što može AGI?
Istraživači smatraju da je matematika granica generativnog razvoja umjetne inteligencije. Trenutačno je generativna umjetna inteligencija dobra u pisanju i prevođenju jezika statističkim predviđanjem sljedeće riječi, a odgovori na isto pitanje mogu uvelike varirati. No, sposobnosti rješavanja matematičkih zadataka, gdje postoji samo jedan točan odgovor, implicira da bi umjetna inteligencija imala veće sposobnosti rasuđivanja nalik ljudskoj inteligenciji. To bi se moglo, vjeruju istraživački AI-ja, primijeniti na i nova znanstvena istraživanja.
Za razliku od kalkulatora koji može riješiti ograničeni broj operacija, AGI može generalizirati, učiti i razumjeti.
U svom pismu odboru, istraživači su naglasili snagu i potencijalnu opasnost umjetne inteligencije, rekli su izvori ne navodeći točne sigurnosne probleme navedene u pismu. Među računalnim znanstvenicima već se dugo raspravlja o opasnosti koju predstavljaju visokointeligentni strojevi ako, primjerice, odluče da im je uništenje čovječanstva u interesu.
Još nije jasno zašto je Altman smijenjen
OpenAI je osnovan kao neprofitna organizacija s Upravnim odborom koji upravlja komercijalnom podružnicom koju vodi Altman. Microsoft je najveći investitor u profitnom poslovanju. Prilikom pregovora oko Altmanova povratka dogovoreno je da će OpenAI imati novi odbor kojim će predsjedati Bret Taylor, bivši izvršni direktor softverske tvrtke Salesforce.
Kreator ChatGPT-ja navodi da je osnovan s ciljem razvoja "sigurne i korisne opće umjetne inteligencije za dobrobit čovječanstva" i da je profitna kompanija "pravno obvezna slijediti misiju neprofitne organizacije".
Naglasak na sigurnosti u neprofitnoj organizaciji doveo je do nagađanja da je Altman otpušten zbog ugrožavanja temeljne misije kompaniji. Međutim, njegov kratkotrajni nasljednik na poziciji privremenog izvršnog direktora Emmett Shear napisao je ovaj tjedan da uprava "nije smijenila Sama zbog bilo kakvog specifičnog neslaganja oko sigurnosti".
https://www.index.hr/vijesti/clanak/sto ... i_ostalo_d
PRIJE nego što je Sam Altman dobio otkaz, OpenAI je navodno radio na naprednom sustavu koji je bio toliko moćan da je izazvao zabrinutost za sigurnost među zaposlenicima kompanije. Model umjetne inteligencije izazvao je takvu uzbunu kod nekih istraživača OpenAI-ja da su se obratili upravnom odboru prije Altmanovog otkaza upozoravajući da bi mogao ugroziti čovječanstvo, prenio je Reuters.
Model nazvan Q* - izgovara se kao "Q-Star" - bio je u stanju riješiti osnovne matematičke probleme s kojima se nikad prije nije susreo, prema stranici s tehnološkim vijestima Information, koja je dodala da je tempo razvoja sustava alarmirao neke sigurnosne stručnjake. "Uz pomoć golemih računalnih resursa, novi je model mogao riješiti određene matematičke probleme", rekao je izvor za Reuters. Sposobnost rješavanja matematičkih problema smatrala bi se značajnim razvojem umjetne inteligencije, piše The Guardian.
Izvještaji su uslijedili nakon višednevnih previranja u OpenAI-ju iz San Francisca, čiji je Upravni odbor otpustio Altmana prošlog petka, ali ga je vratio na posao u utorak navečer nakon što je gotovo svih 750 zaposlenika kompanije zaprijetilo odlaskom ako se ne vrati. Altman je također imao podršku najvećeg investitora OpenAI-ja - Microsofta.
Razvoj opće umjetne inteligencije?
Mnogi su stručnjaci zabrinuti da se kompanije kao što je OpenAI kreću prebrzo prema razvoju opće umjetne inteligencije (AGI), odnosno prema sustavu koji može obavljati širok raspon zadataka na ljudskoj razini ili iznad razine ljudske inteligencije - i koji bi u teoriji mogao izbjeći ljudsku kontrolu.
Andrew Rogoyski s Instituta za umjetnu inteligenciju usmjerenu na ljude na Sveučilištu Surrey, rekao je da bi postojanje velikog jezičnog modela (LLM) za rješavanje matematike bio napredak: "Intrinzična sposobnost LLM-a da rješava matematiku veliki je korak naprijed koji dopušta umjetnoj inteligenciji posve novi niz analitičkih sposobnosti".
Govoreći u četvrtak prošlog tjedna, dan prije iznenadnog otkaza, Altman je naznačio da je kompanija koja stoji iza ChatGPT-ja napravila još jedan iskorak. U govoru na summitu Azijsko-pacifičke gospodarske suradnje (APEC), rekao je sljedeće: "Četiri puta u povijesti OpenAI-ja, a posljednji put prije tek nekoliko tjedana, bio sam u prostoriji kada je na neki način skinut veo neznanja i granica otkrića pomaknuta naprijed. Profesionalna je čast sudjelovati u tome."
Što može AGI?
Istraživači smatraju da je matematika granica generativnog razvoja umjetne inteligencije. Trenutačno je generativna umjetna inteligencija dobra u pisanju i prevođenju jezika statističkim predviđanjem sljedeće riječi, a odgovori na isto pitanje mogu uvelike varirati. No, sposobnosti rješavanja matematičkih zadataka, gdje postoji samo jedan točan odgovor, implicira da bi umjetna inteligencija imala veće sposobnosti rasuđivanja nalik ljudskoj inteligenciji. To bi se moglo, vjeruju istraživački AI-ja, primijeniti na i nova znanstvena istraživanja.
Za razliku od kalkulatora koji može riješiti ograničeni broj operacija, AGI može generalizirati, učiti i razumjeti.
U svom pismu odboru, istraživači su naglasili snagu i potencijalnu opasnost umjetne inteligencije, rekli su izvori ne navodeći točne sigurnosne probleme navedene u pismu. Među računalnim znanstvenicima već se dugo raspravlja o opasnosti koju predstavljaju visokointeligentni strojevi ako, primjerice, odluče da im je uništenje čovječanstva u interesu.
Još nije jasno zašto je Altman smijenjen
OpenAI je osnovan kao neprofitna organizacija s Upravnim odborom koji upravlja komercijalnom podružnicom koju vodi Altman. Microsoft je najveći investitor u profitnom poslovanju. Prilikom pregovora oko Altmanova povratka dogovoreno je da će OpenAI imati novi odbor kojim će predsjedati Bret Taylor, bivši izvršni direktor softverske tvrtke Salesforce.
Kreator ChatGPT-ja navodi da je osnovan s ciljem razvoja "sigurne i korisne opće umjetne inteligencije za dobrobit čovječanstva" i da je profitna kompanija "pravno obvezna slijediti misiju neprofitne organizacije".
Naglasak na sigurnosti u neprofitnoj organizaciji doveo je do nagađanja da je Altman otpušten zbog ugrožavanja temeljne misije kompaniji. Međutim, njegov kratkotrajni nasljednik na poziciji privremenog izvršnog direktora Emmett Shear napisao je ovaj tjedan da uprava "nije smijenila Sama zbog bilo kakvog specifičnog neslaganja oko sigurnosti".
Najgore je svađati se sa budalom. Prvo te spusti na svoj nivo, a onda te dotuče iskustvom...
-
Gazu
- Postovi: 2110
- Pridružio se: 08 Feb 2022, 00:31
- Garaža: Opel Astra G 1.6i-16V
- Lokacija: Novi Sad
Re: Nauka i tehnologija
Niti mogu da generalizuju, niti da uče, a pogotovo ne da razumeju. Večito iste gluposti.
Uhvatio sam se samo za ovo jer sam čitao tekst na preskok, sigurno ima još dosta neistina.
Re: Nauka i tehnologija
Jesi li cetovao nekad sa AI? U principu 1. nivo korisnicke podrske lagano moze da se zameni ovim, plus je strucnost neretko veca, ne placas kancelarije, ne radis obuke kad ode radnik... Gugl prevodilac je pre desetak godina bio neupotrebljiv, a pogledaj sad. Veliki broj poslova ce biti ugasen u narednim decenijama. Njihov plan je izmedju ostalog i da kamioni ubuduce idu potpuno samostalno.
Послато са SM-N960F помоћу Тапатока
Послато са SM-N960F помоћу Тапатока
-
Gazu
- Postovi: 2110
- Pridružio se: 08 Feb 2022, 00:31
- Garaža: Opel Astra G 1.6i-16V
- Lokacija: Novi Sad
Re: Nauka i tehnologija
Jesam, i bio mi je super koristan za moje potrebe. Fantastičan alat. Ne razume on šta mi piše, i šta mu ja pišem. Princip funkcionisanja je drastično različit od čoveka.
Ali on ne ume da generalizuje, ne ume da uči, ne može da razume. U potpunosti se slažem sa tobom da ovo može da zameni potrebu za raznoraznim Indijcima, kamiondžijama, i slično, ali to ne govori o inteligenciji "veštačke inteligencije", već o prirodi tih poslova (a pod time - ne mislim ništa loše o ljudima koji se tim poslovima bave). To su poslovi koje može da zameni mašina - koja će to obavljati bez ikakve svesti ili razumevanja toga što radi. Skoro mehanički posao.
Da će ovo imati veoma gadne posledice po društvo, da, hoće!
Ali on ne ume da generalizuje, ne ume da uči, ne može da razume. U potpunosti se slažem sa tobom da ovo može da zameni potrebu za raznoraznim Indijcima, kamiondžijama, i slično, ali to ne govori o inteligenciji "veštačke inteligencije", već o prirodi tih poslova (a pod time - ne mislim ništa loše o ljudima koji se tim poslovima bave). To su poslovi koje može da zameni mašina - koja će to obavljati bez ikakve svesti ili razumevanja toga što radi. Skoro mehanički posao.
Da će ovo imati veoma gadne posledice po društvo, da, hoće!
-
grujicd
- Postovi: 8365
- Pridružio se: 16 Jan 2012, 13:32
- Garaža: Honda Jazz 1.4 2007 & CR-V 2.0 2012
- Lokacija: Beograd
Re: Nauka i tehnologija
AGI je cilj, ali trenutni AI je daleko od toga. Te najave da "tek što nije" mi ne piju vodu, ovo što sad koristimo jeste i korisno i impresivno ali nije ni blizu pravog razumevanja.
Da sam kamiondžija ne bi se preterano brinuo, dosta je često mišljenje da potpuna autonomna vožnja ne može bez potpunog AGI, tj. inteligencije na nivou čovekove. Koliko god da autonomni sistemi napreduju jako je teško pokriti sve granične slučajeve ako sistem ne razume celokupno okruženje na istom nivou kako ga i čovek razume. Ako naiđe nepredviđena situacija samo može da stane, bez improvizacije.
