Making sense of the world through data
The focus of this blog is #data #bigdata #dataanalytics #privacy #digitalmarketing #AI #artificialintelligence #ML #GIS #datavisualization and many other aspects, fields and applications of data
Yesterday, I introduced some fundamental questions about the cosmos which bring doubts about our current view of the Universe. In spite of General Relativity and Quantum Theory's successes, there are still many questions marks about our understanding of the cosmos, some deep enough to oblige us to completely rethink our theories. We may be on the edge of a new paradigm of physics.
Today, it is the turn of evolution: the great theory of Darwin. The purpose of the exercise is not to take us back to creationism but to progress towards a more complex and fundamental understanding of evolution. From the beginning, it was understood that "Darwin" was not enough to explain evolution so in the later part of the 20th Century came the concept of punctuated evolution. The idea that species are mostly static but that at some stage they go through an accelerated patch of evolution thanks to an "arm race" introduced by new factors or conditions. The best example is probably the sudden growth of our brains thanks to the discovery of abstract language. Then came the discovery of Epigenetics with the understanding that methylation is a way for the environment to actually have a direct effect on our gene. Could it be that Lamark was not completely wrong in the end?
But the discussion below goes one step further: Could evolution be directed? This sound incredible and would most certainly be anathema to die hard Darwinists since such a theory seems to contain a whiff of sulfuric "Creationism" but this is not what it is. As for the cosmos yesterday, some questions of evolution do not seem to make sense. One in particular is extremely hard to answer: If evolution was truly blind and completely directionless, how on earth was it so fast? The possibilities of combining genes to create new "things" of functions are truly infinite so to test them all randomly and eliminate the bad ones would take forever which is clearly not what we see. Evolution takes a long time but not forever. So how do we explain this? The short answer is that we do not! As for cosmology, we just ignore this pesky Question.
There is much more in the long discussion below. This is true science. No the "science" of semi-literate European bureaucrats whose science is whatever is convenient for their purpose as we saw during Covid. But real science of hard questions and experimentation to get closer to the truth. It is difficult, messy, mind wrecking, full of turns, cul-de-sacs and wrong paths but in the end, exhilarating.
Cosmology has a problem! The Universe doesn't behave as we expected. In particular, galaxies do not rotate according to Newtonian principles! This is huge because it shows that we understand much less than we thought we did. We invented "Dark Matter" to deal with this inconvenience but in the end, it looks more and more like Dark Matter will join the Ether in the black hole of discarded ideas. This is both a disappointment and an exhilarating time to be alive. (see below)
But this is not all. The James Webb telescope is on the verge of finding fully formed galaxies far too close to the origin of the Universe. They should not have had enough time to form. The days of the Big Bang theory may be numbered. But in reality they already were long ago. It has been a well known secret that some stars look uncannily older than the Universe by at least one and sometimes two billion years or more. How could this be? I remember a time not so long ago when we expected the Universe to be at least 15 billion years old because that was the estimated age of the oldest red dwarf stars. So it came as a shock when suddenly the consensus based on the calculation of the cosmic constant settled on 13.7 billion years. That was simply not enough time! We had a Benjamin Button problem! Scientists have been trying to fill this gap for the last 20 years to no avail.
Einstein once said about the Universe that the most extraordinary thing was that it was understandable. But is it? Or are our theories just weak approximations of a strangely complex underlying reality?
In his classic book On the Structure of Scientific Revolutions,
the philosopher Thomas Kuhn posited that, for a new scientific
framework to take root, there has to be evidence that doesn’t sit well
within the existing framework. For over a century now, Einstein’s theory
of relativity and gravity has been the existing framework. However,
cracks are starting to show, and a new paper from researchers at Case
Western Reserve University added another one recently when they failed
to find decreasing rotational energy in galaxies even millions of light
years away from the galaxy’s center.
Galaxies are known to rotate – even our solar system travels in a
circle around the center of the Milky Way galaxy at around 200 km per
second, though we can’t perceive any motion on human time scales.
According to Newtonian dynamics, this rotational speed should slow down
the farther away a star is from the center of a galaxy. However,
observations didn’t support this, showing that the speed kept up no
matter how far away the star is.
That led scientists to create another force impacting the speed of
rotation of the farthest-out stars. Today, we commonly call it dark
matter. However, scientists have also spent decades trying to puzzle out
what exactly dark matter is made of and have yet to come up with a
coherent theory.
Anton dives into a weird quirk of galaxy rotation. Credit – Anton Petrov YouTube Channel
But in some cases, even the existence of dark matter as we know it
doesn’t match the observational data. Dr. Tobias Mistele, a post-doc at
Case, found that the rotational speed of galaxies doesn’t drop off, no
matter how far out they are and no matter how long they’ve been doing
so. This data flies in the face of a traditional understanding of dark
matter, where its gravitational influence is felt by a “halo”
surrounding the dark matter itself. Even these dark matter halos have an
effective area. Dr. Mistele and his co-authors found evidence of
maintained rotational speed that should be well outside the sphere of
influence of any dark matter halo existing in these galaxies.
To collect this data, the authors used a favorite tool of
cosmologists – gravitational lensing. They collected data on galaxies
that were far away and had their light amplified by a galaxy cluster or
similarly massive object that was nearer. When collecting the data, Dr.
Mistele analyzed the speed of rotation of the stars in a galaxy and
plotted it against the distance of those stars from the galaxy’s center.
This is known as a “Tully-Fisher” relation in cosmology.
The result was an almost perfectly straight line – the rotational
speed of stars in a galaxy did not seem to diminish with distance from
the galaxy’s center, as both traditional Newtonian dynamics and
relativity via dark matter predicted it would. So, what alternative
explanations are there?
Why do galaxy rotation curves matter? Nora explains. Credit – Nora’s Guide to the Galaxy YouTube Channel
Paper co-author Stacy McGaugh points out in a press release that one
theory in physics accurately predicted the data his team had
collected—the modified Newtonian Dynamics (or MOND) theory. Designed
explicitly to account for things like galaxy rotations, MOND was
developed in 1983 and remains controversial to this day. It struggles
with things like the gravitational lensing with which the paper’s data
was collected.
That disconnect points to the need for a deeper understanding of
gravity – what Kuhn called a “crisis,” which many cosmologists already
believe is afflicting the discipline. While there is no current
consensus on what might resolve that crisis, the evidence is mounting
for the need for resolution. If we’re truly going to understand our
place in the universe, we will eventually need to figure out a solution –
it just might take a while.
I have already written too much on the failure of the "green transition" in Germany so I won't add a long monologue here. Just to say that this was predictable. The Germans did put the EV cart in front of the transition bull and will soon find them selves in the very unenviable position of de-industrializing while having to choose between cheap Chinese EV and much higher inflation. They should have listened to the advise of Toyota's President and invested first into hybrids and cleaner ICE vehicles waiting a little longer for better electric technologies to emerge. Now the risk is that their automotive industry will be stuck between a rock (the law) and a hard place (Chinese competition) with nowhere to go. What's the real market for 100,000+ Euros cars and isn't that market already occupied by Tesla?
30 years of decline already and the fall is still going strong. If you live in Tokyo, within the Yamanote train line you'll probably notice nothing at all. The bubble of the 1980s has been replaced by a new one. Speculation is "healthy" (if that's the right word!) with hundreds of towers being built yearly to the point of transforming the city into a forest of high-rise buildings. Who will work or live there is of no importance. The yen is cheap so relatively easy to invest in. Public deficits and carry trade help create tons of money which somehow has to go somewhere. Luckily or not, depends on your point of view, not into inflation since salaries are kept under control by fierce competition.
Likewise, the debt is mostly domestic, or rather was, since by now most of the JGB (Japan Government Bonds) have been bought by the central bank to the point that trading is almost completely gone! So Japan could (and actually does) go down year after year into the abyss (which strangely sits not too far from the country geographically.)
But stray away from central Tokyo and under the water, all is not well. Japan is becoming poor, literally and practically. Here's one obvious symptom beyond the empty cities and shrinking birth rates: The decline of fish consumption which is crashing thanks to exploding prices.
TOKYO, Jun 18 (News On Japan)
-
Fish consumption in Japan has hit a record low, with annual per capita
intake dropping from about 40 kilograms in 2001 to roughly 22 kilograms
in 2022. This decline is particularly notable among seniors, who are
increasingly opting for meat over fish.
Factors contributing to this trend include family preferences for
meat, high fish prices, and the inconvenience of fish preparation.
Local fish shops' disappearance also plays a role.
But here's another more insidious factor: As the yen keeps falling and resentment grows, Japan, like other developing countries (Thailand comes to mind) is brain-storming how to milk foreigner tourists for all they're worth.
Himeji Castle, a popular tourist destination and UNESCO World Heritage
site, is currently considering a significant increase in admission fees
for foreign visitors. The admission fee for adults is presently 1,000
yen, but the mayor of Himeji City has proposed raising the fee to 30
dollars for foreign tourists, a move that has sparked considerable
debate.
It is said that beggars have no shame since they can't afford it. So how far down the path of bankruptcy Japan really is? Hard to say. For countries as for people, as Hemingway once said, bankruptcy comes very slowly at first then suddenly. Left to its own demise, Japan could likely go on another decade on this road to Zen nothingness and probable misery. More likely, international events will decide differently sooner than that!
Japan couldn't find the exit to its real estate bubble and neither is China now. The simple reason is the scale of the bubble, usually many times that of the economy.
So what comes next for China? The 1990s in Japan were stressful financially but not yet a catastrophe. Huge projects were being completed just as in China today, while more debt was piling up. The problem was floundering growth, again just as China today and the fact that most new projects were uneconomical and would therefore bring no future growth. (Imagine what was going to happen to Olympic swimming pools in villages which didn't have the money to pay for the upkeep?)
This doesn't mean quite yet the end of China but as for economic growth, 10% certainly belongs to the past. Now money will be more scarce and investment will have to be more selective. This is where Japan failed, completely missing the digital revolution, while investing in absurd projects such as the computer 5th generation, whatever it meant for people then, which brought absolutely nothing. Hopefully China will be wiser although "wise" has its limits which is why it is often better to rely on market forces.
So much for that Chinese housing "bailout", which we correctly warned would be woefully insufficient.
A
slew of data published early Monday local time, showed that among
various other economic measures, China’s housing slump deepened in May
and triggered new calls for the government to pump cash and credit into
the economy, while industrial output, which has kept growth on track,
fell short of forecasts.
New home prices - the most important
indicator of middle-class wealth in the world's second biggest economy -
dropped at the fastest pace since October 2014, falling 0.7% m/m in May
(v/s -0.58% in April) and marking the 11th straight decline despite the government’s stimulus to support the property market.
On an annual basis, home prices slumped 4.3%, the biggest drop since
the summer of 2015. In fact, the last time home prices plunged so much
Beijing pursued a massive yuan devaluation that led to a $1 trillion in
fx outflows to stabilize and sparked the very first mega meltup in
bitcoin which sent it from $200 to far over $1000.
Staying
on China retail sales rose +3.7% y/y in May, exceeding market
expectations for a +3.0% gain and increasing pace from a +2.3% increase
in the previous month but Chinese shoppers remain far from recovering
their pre-pandemic mojo. However, other economic metrics failed to
surpass market forecasts with industrial output growing +5.6% y/y in May
(v/s +6.2% expected), down from an increase of +6.7% in April and
missing the median forecast in a Bloomberg survey.
Meanwhile,
the nation’s real estate crisis continued to weigh on investment in
fixed assets with the overall YTD investment figures expanding +4.0%,
also below estimates of +4.2% gain.
“The
most disappointing in May’s data is probably that property sales barely
saw any improvements even after so many supportive measures,”
said Jacqueline Rong, chief China economist at BNP Paribas SA. She said
China’s authorities need to find ways to lower the rates on existing
mortgages, closing the gap with the cost of new ones.
The numbers -
excluding the dire real estate prints - add up to a still-weak recovery
most economists said, and will likely require more action from Beijing
to bolster consumer demand and tackle imbalances, if this year’s 5%
growth target is to be met. That could take the form of stepped-up
government spending and heightened efforts by the central bank to put a
floor under housing markets and get credit flowing.
On the other
hand, if one includes China's collapsing property market - which once
upon a time was the world's largest asset class - it becomes clear that
while Beijing may pretend it does not need a huge fiscal and monetary
stimulus, it's
Late last month, China unveiled a broad
rescue package to prop up housing sales as a credit crisis was engulfing
some of the country’s biggest real estate developers. It relaxed
mortgage rules and encouraged local governments to buy unsold homes.
However, as we warned and
many others agreed, the financial incentives aren’t big enough and
trial programs in several cities have shown progress can be slow.
Subdued
demand at home and the deteriorating foreign-trade environment are
weighing on business confidence, discouraging companies from investing
and driving some to move production overseas. Credit growth has
been lackluster and the M1 money supply gauge contracted in May at the
fastest rate in data going back to 1996.
Elsewhere, the
PBOC on Monday kept a key interest rate unchanged for the
tenth straight month. Economists say the bank’s room to cut rates is
constrained by the need to prop up the yuan, which faces downward
pressure as the US Federal Reserve reinforces its high-for-longer
message.
China’s growth remains “highly uneven, with exports and
new energy-related capex as the drivers while consumption and property
as the drags,” according to economists including Larry Hu at Macquarie
Capital Ltd. Still, the slowdown isn’t severe enough to threaten the
growth target and while policymakers may take some limited action “the
urgency for a major stimulus is low,” they wrote
In a survey of
more than 400 top executives conducted by UBS Group AG over roughly a
month through mid-May, firms reported weaker prospects for orders,
revenue and margins compared with the same period of 2023. There was a
drop in the share of respondents who plan to increase capital
expenditure in the second half of this year.
“We still need to see
new stimulus coming in,” said Helen Qiao, chief Greater China economist
at Bank of America Global Research, in a Bloomberg TV interview.
“Otherwise the growth momentum could very much weaken.”
Often we're talking about the "coming" crisis as if it was something abstract which sometimes in the future will strike us out of the blue. Well, think again or more practically, read The Great Taking.
The current financial crisis has been in the making for the last 50 years, since at least the "temporary" decoupling of the US dollar and gold in 1971. Now the money you have on your bank account is a lot of things before being yours. In the end you're just a lowly creditor to your bank. So that when the system goes under, and it will eventually, the money will shore up the bank first and what is left will be yours. Nominally at least since practically, you probably won't have access to it.
There is much more in the video below, so well worth listening to. (The alternative is spending a few weeks in Argentina in order to understand how a post bankrupt country operates. Since it is not very fun, most people won't do it. Quite a shame.)
Second part of the video when they're "bullish" about the US is short and can be listened to at x2 speed. I guess you've got to say something positive after you spill the financial beans!
If you enjoyed the first video, here's another one, just as good.
Using Japan to strike at India to harm Russia? Sound machiavelian but it might also be stupid.
Indians have no love for China but it doesn't prevent them from working with the Chinese through the BRICS. This means that the priority for India at this stage is economic beyond anything else. Seen through such a lens, what does a declining Japan offers which would be worth jeopardizing trade with a resurgent Russia?
If the answer is "nothing" then Japan will be shooting itself in the foot. If at least the hope that Japan can transfer some technologies as it did to China 20 years ago was present, it could be a mitigating argument. But the chance at this stage is almost zero. This means that Japan will have nothing to offer. Just a stick with no carrot. This is not exactly a great position to start a negotiation!
India
might conclude that Japan was put up to this by its American patron if
it comes to pass, which could then complicate their multilateral efforts
to manage China’s rise.
Tribune India cited
Japanese media to report that Chief Cabinet Secretary Yoshimasa Hayashi
recently said that “At the recent G-7 summit, we have announced that we
are considering a new package of sanctions that will include companies
from third countries. We are looking at measures against companies from
China, India, the UAE and Uzbekistan.” The Economic Times confirmed this report in their article citing unnamed sources who are allegedly familiar with the situation.
It
would be a bad idea for Japan to sanction Indian companies on
anti-Russian pretexts since that would toxify their strategic ties. India and Japan cooperate in the Quad, which is nowadays mostly just a talking club unlike former
Russian Defense Minister Sergey Shoigu’s claim that it’s a
US-controlled anti-Chinese proxy group, and are also close economic
partners. Their hard-earned mutual trust that was built in the years
after Japan sanctioned India in 1998 for its nuclear tests would be instantly shattered.
The consequences of that happening could complicate the US’ grand strategic plans in Asia, which are partially predicated upon the comprehensive strengthening of Indo-Japanese relations. India’s problems with China are independent of the US’ or Japan’s, but it finds common cause with them in managing that country’s rise. Nevertheless, the newfound trouble in
its ties with the US coupled with the scenario of Japan sanctioning its
companies could impede their multilateral cooperation in this respect.
India
might also conclude that Japan was put up to this by its American
patron as part of the latter’s plans to pile more pressure upon it
for the reasons that were explained in the preceding hyperlinked
analysis, which can be summarized as punishing it for refusing to dump
Russia. After all, the US could always intervene behind the scenes to
stop Japan if it was really concerned about the consequences that these
sanctions could have on their cooperation with India vis-Ã -vis China,
but it might not do so.
In that event, American policymakers would
have calculated that it’s more important to punish India than to
continue working with it in furtherance of their common cause, which
would in turn suggest that more pressure might then be applied against
it in the future on anti-Russian or other pretexts. The US’
liberal-globalist faction interpreted the latest general elections in
India as weakening Prime Minister Narendra Modi so it’s possible that
they’re emboldened by that to ramp up their pressure against him.
Instead
of doing so directly, they might have decided to first act through
Japan via the means of having that country sanction its companies for
doing business with Russia, after which it can’t be ruled out that other
G7 states such as the US itself might then follow its lead as part of a
preplanned policy. To be clear, that also might not happen at all, with
or without Japan doing what Hayashi just announced. Even so, however,
it’s credible enough of a possibility for India to think about just in
case.
Delhi would have to respond if this sequence of events
unfolds, though it might only take the form of harsh denouncements as
opposed to any meaningful response owing to India’s complex economic
interdependence with the G7, which also serves to limit the extent of
the bloc’s potential sanctions too. In any case, bilateral trust would be broken, and India might defiantly double down even further on its relations with Russia in order to send the message that it won’t be deterred by such pressure.
The fundamental question: "Where do we come from?" has now been redefined as: "Are we living in a simulation?" and the answer based on what we know is probably "Yes!" But this answers nothing. Who created the simulation and where do they come from?
I like the geometrical theory explained below. We know from general relativity and quantum physics that geometric principles underlie our reality so it is probably the right approach to try to grasp what this "reality" is made of.
To me, this is "real enough" even though it may be virtual. As the virtual character said in "13 Floor" (A must see movie on the subject) when shooting at the hero with a gun: "Is this real?"
We may be living in a virtual universe but luckily, the beings who devised the "box" (our universe) did make it large enough for us to grow and prosper inside. Although it says something profound about our nature, that barely out of the forest, we're already lurking beyond the sphere of the known universe!
Once in a while it is worth taking a break and just looking at the world around us. (All the better if it is through the cockpit of a jetliner. :-)
Having had the chance to witness such beauty around the Pacific, I can attest that thunderstorms at night in the tropics compete with auroras at the Poles for a magic show of pure beauty and colors. Likewise, flying low over coral reefs surrounding atolls gives us a multi-color view of the largest mega-organisms on Earth. Add a frequent rainbow after the rain or a lava trail in Hawaii, Vanuatu or New Britain and it is easy to understand why so many people believe that the Earth is alive.
A former software systems engineer, Mr. Borja is from Ecuador and flies a Boeing 767 for a major airline in the region.
Before becoming a storm photographer,
Mr. Borja was interested in photography as a hobby and used an old film
camera—which he inherited from his father—before moving to a digital
SLR.
“Storms always fascinated me, and I kept wondering how I could capture such beautiful and amazing phenomena,”
he said. “I tried different settings and techniques until I finally
came up with a way to consistently capture storms from a moving
airplane.”
Mr.
Borja, who typically flies between 30,000 and 40,000 feet above the
earth, often encounters cumulonimbus storms that vary in intensity.
“I would say it is very rare when we don’t encounter storms,” he said. “Almost every flight there is some moderate storm activity in our surroundings.”
As
an airline pilot, Mr. Borja is unable to modify his route, so he makes
use of the various opportunities to capture a good image.
“If the
airplane is too close to the storm or the cloud system, the turbulence
and the clouds themselves make it impossible,” he said. “I need calm air to be able to capture these storms.”
Many have wondered how Mr. Borja can fly and capture such incredible shots at the same time.
To
which he replies that since he often flies transatlantic routes, the
three other pilots with him take turns controlling the aircraft, leaving
him four hours of rest time to sleep and take pictures.
The pilot is unable to carry a lot of photography gear with him at all times, so he almost always uses his full-frame DSLR with a 28-300 mm lens.
“For storms, I don’t really need a long lens, but this is my lens for any occasion,” he said.
His
proudest accomplishment has been capturing “Pacific Storm,” a photo
that was taken over the Pacific Ocean in 2004. This image was recognized
by National Geographic and widely viewed by scientists worldwide.
“I
ended up learning a lot about meteorology and science by talking to all
the people that got interested in this image,” Mr. Borja said. This
success allowed him to publish his book—“#The Storm Pilot.”
Storm photography “takes a lot of luck and a lot of trial and error,” according
to Mr. Borja. Since the cockpit is bordered by high-quality glass
windows that don’t produce much glare, this becomes an advantage for
him. However, other variables such as light are beyond his control.
“The
more light, the more difficult [it is] to capture a storm,” he said.
“Sometimes with a full moon or intense city lights, the scenery is
fantastic but very challenging to capture.”
The humble storm photographer has always been willing to learn more about his craft.
“I
have greatly expanded my photographic knowledge and practice thanks to
the people I’ve met through this journey,“ Mr. Borja said. ”I’ve had the
opportunity to talk to great photographers, who have taught me some
great lessons.”
He
advises up-and-coming storm photographers to learn the skills of
photography well but not to take those guidelines as rigid rules, or
they risk limiting their creativity.
“I was taught that you cannot
take a long exposure from a moving airplane with a handheld camera, and
yet, I managed to come up with a strategy to make it work,” he said.
In sharing his spectacular images with the world, he wants people to enjoy nature and all it has to offer.
“We live more and more inside our human-made structures and environments,“ Mr. Borja said. ”I deeply enjoy flying and looking at our planet from a distance where we humans are imperceptible.
“Too often we forget to look up and enjoy the awesome views that nature is giving us for free.”
There is a great line in the Terminator movie when Schwarzenegger drops a guy saying "I lied!". profoundly prophetic.
For years, AI scientists told us there was nothing to worry about AI since the systems would be thoroughly trained and aligned on human values before being released. What a joke! There people were either imbeciles (which knowing some of them, I know they're not!) or plainly deceiving us (and probably themselves!)
The gist of the problem is that we now understand that intelligence is an emergent property made of many characteristics. But in what order do these characteristics emerge is still a complete mystery. It seems clear now that when winning is the goal, lying is a winning strategy and deceiving an even better one.
This is a multi-dimensional problem, among those well known not to have a solution. You can control some of the factors but not all of them. It is simply too complex. We will eventually stumble into the creation of true monsters about which We'll have no idea what to do. With or without consciousness, these machines will outsmart us a thousand times at a million times the speed. Very soon they will instantly map out the universe of our human possible reaction to each of their action. By that time, it will be check and mate! Game over!
A new study has found that AI systems known as large language models (LLMs) can exhibit "Machiavellianism," or intentional and amoral manipulativeness, which can then lead to deceptive behavior.
The study authored by German AI ethicist Thilo Hagendorff of the University of Stuttgart, and published in PNAS, notes that OpenAI's GPT-4 demonstrated deceptive behavior
in 99.2% of simple test scenarios. Hagendorff qualified various
"maladaptive" traits in 10 different LLMs, most of which are within the
GPT family, according to Futurism.
In another study published in Patterns found that Meta's LLM had no problem lying to get ahead of its human competitors.
Billed as a human-level champion
in the political strategy board game "Diplomacy," Meta's Cicero model
was the subject of the Patterns study. As the disparate research group —
comprised of a physicist, a philosopher, and two AI safety experts —
found, the LLM got ahead of its human competitors by, in a word, fibbing.
Led by Massachusetts Institute of Technology postdoctoral researcher Peter Park, that paper found that Cicero not only excels at deception, but seems to have learned how to lie the more it gets used — a state of affairs "much closer to explicit manipulation" than, say, AI's propensity for hallucination, in which models confidently assert the wrong answers accidentally. -Futurism
While Hagendorff suggests that LLM deception and lying is confounded by an AI's inability to have human "intention," the Patterns study calls out the LLM for breaking its promise never to "intentionally backstab" its allies - as it "engages in premeditated deception, breaks the deals to which it had agreed, and tells outright falsehoods."
As Park explained in a press release, "We found that Meta’s AI had learned to be a master of deception."
"While Meta succeeded in training its AI to win in the game of Diplomacy, Meta failed to train its AI to win honestly."
Meta replied to a statement by the NY Post, saying that "the models our researchers built are trained solely to play the game Diplomacy."
Well-known for expressly allowing lying, Diplomacy has jokingly been referred to as a friendship-ending game
because it encourages pulling one over on opponents, and if Cicero was
trained exclusively on its rulebook, then it was essentially trained to
lie.
Reading between the lines, neither study has
demonstrated that AI models are lying over their own volition, but
instead doing so because they've either been trained or jailbroken to do
so.
And as Futurism notes - this is good news for those concerned about AIs becoming sentient anytime soon - but very bad if one is worried about LLMs designed with mass manipulation in mind.
