Construction is expected to begin this spring on the University of Texas’ newest field lab. The lab is being funded with a $50 million grant, the largest donation in

New UT Hill Country lab will soon pop up near ‘best-kept secret’ in Texas

Justin Wood spoke with Art Director Forrest Schehl about the new Secret Lair X 'Evil Dead' release and how he designed the new set of cards.

‘Evil Dead’ Comes To The Magic: The Gathering Universe [Exclusive Interview]

Your smartwatch could be the key to better, more personalized cancer care, UVA Cancer Center researchers suggest. A team of UVA Health researchers has demonstrated the potential of wearable and mobile devices such as smartwatches and smartphones to help doctors tailor treatments to the needs of individual patients.

The secret to better cancer care could already be on your wrist

Nonetheless, once you've unlocked the door make your way down into Connors's secret lab. Now inside you'll have to snoop around for some surveillance footage. This will reveal what happened in the lab and progress you to the next point in the mission.

Marvel’s Spider-Man 2: How to Solve The Piano Puzzle (Conner’s Secret Door)

The World Series, like the regular season, will use baseball mud that "spreads like cream and grips like sandpaper."

What’s so ‘magic’ about the secret South Jersey mud rubbed on baseballs? These Penn researchers think they know why.

Matthew Connelly, who runs Columbia’s History Lab, researches the answer to that and other questions about U.S. government records.

Can You Keep a Secret?

In order to find the four-digit code for the piano puzzle at Dr. Connors' house, Marvel's Spider-Man 2 players will need the help of a child's toy.

What's the Dr. Connors' Piano Code in Marvel's Spider-Man 2?

I’ve been taking lion’s mane mushroom for a while, and it really helps me focus and boost my energy,” said Gisele Bündchen. “I love it!”

Gisele Bündchen’s secret energy booster: What is lion’s mane mushroom and what are its benefits?

Your outfits should be chosen based on your eye colour, a study suggests. Previously, experts thought skin tone was the most important factor when determining if the colour of a garment would suit the wearer,

Secret of picking the perfect outfit colour is revealed

A Chinese-linked company was found running an unlicensed California biolab that contained at least 20 potentially infectious diseases, including coronavirus, HIV, hepatitis, and herpes. Jia Bei Zhu, a Chinese citizen residing in Clovis,

Behind A Secret Chinese Biolab in California

Secret Lab

secretlab

In 2021, NFTs have blown up in a big way, so everyone is scrambling to produce their own non-fungibles. This results in an oversaturated market, where buyers are left with too much choice, and artists are desperate to make their work stand out from the crowd. Creating GORECATS, our first collection of NFTs proved to be an interesting challenge for us. With more than 20 years of character design and direction under our belt, — including huge names like Marvel, Disney, Amazon Studios & Nickelodeon, —  Jim Bryson, our character designer and I wanted to inject all that experience into a really compelling NFT collection. We learned we’d need three things to succeed: a high-quality art-driven project with a hook, a strong team, and an engaging story. Without further ado, here’s the story of the Gorecats. I hope it proves inspiring — and useful — for all talented NFT creators-to-be, who can draw from our experience as professional IP creators and storytellers. Capture a feeling One of the main reasons why we buy things is because they make us feel a certain way. Joyful. Successful. Secure. However, from our experience, few things can compare to the pull of nostalgia. This is why our NFT collection had to feel like it was part of something bigger — similar to Funko POP! figurines, where you would instantly recognize the characters, and therefore want to collect them all. It’s important, though, to be genuine, and use what resonates with you. Jim and I both grew up watching B-movie horror films like Evil Dead, Bad Taste, Child’s Play, and Gremlins. This is how we came up with the idea to capture that love of pop culture, 80s horror movies — and, of course, cute pixel art cats — into one collection. Testing the waters & pulling the crew Every idea may be a hit or a miss. Before you embark on a project, it’s vital to know what your audience wants, so, with all our products, we always test the market first. Last summer, Jim posted some gruesome cat sketches on a ‘Gory-Garfield’ thread on Reddit. When we got a huge response, we knew we really struck a chord with the fans. So Jim and I spent the next 3 months developing the designs into the bold, bizarre, and extremely bloody GORECATS collection. ‘You never go far alone’, — next, we needed a strong team. Before GORECATS, we had been working in the NFT space for 6 months, producing artwork for tokens like X-cad and Blockasset, so we used our contacts to connect with talented developers, tokenomists, puzzle designers, and marketers. Even though it’s common to remain anonymous in this industry, we felt it was good to be very transparent about who was behind the project. The power of storytelling When you introduce a story behind the art, art gains meaning. GORECATS was meant to be much more than just a generative collection of illustrated cats. First, we made sure the characters are relatable. Each Gorecat has a unique personality, our fans can identify and connect with. Second, we wanted to deeply engage the audience by playing on their curiosity, so we developed a mystery backstory about a detective who discovers a series of gruesome cat murders. By joining our discord group, fans will be able to interact with the story, solve clues, unlock hidden rewards — and, ultimately, discover the true killer that is hidden somewhere in the collection. More utility, better UX Our next move was to create more fun utilities for our NFT buyers, so our plot has an interesting twist: the detective discovers a secret lab that has been experimenting on Gorecats, breeding them and genetically altering their DNA. When the experiments go awry, one of the Gorecats in the lab gets infected with a deadly biological virus, and, of course, escapes. The virus spreads through the breed, turning them into relentlessly aggressive zombie cats. Building a brand So, you’ve created your NFT collection with quality visuals and a compelling story, and it is proving to be a success. Why stop there? Convert it into a brand. By doing so, you will turn your NFT buyers into loyal followers, eager to keep an eye on new products, see what else is in store, and follow the larger storyline. We’ve already done this back in 2013 when we saw vast storytelling potential in the iPad and produced the first fully animated comic called ‘Niko & the Sword of Light’. The project hit #1 on the AppStore and then went on to become an Emmy-winning TV series on Amazon Prime. Now we have the same goal in mind for GORECATS. The plan is to replicate the model we used with ‘Niko’ and develop it from a collection of NFTs and a mystery comic into an animated series – and on to a full-blast universe inhabited by the Gorecats and powered by their stories. Summing up the gore Having spent quite a lot of time in the fast-moving, highly competitive NFT market, we’ve created some recipes for creating a collection that really stands out. Here are some of the key ingredients: a unique hook or angle, compelling characters with an engaging backstory, high-quality art, a strong team — and, above all, a clear marketing plan. Sadly, we’ve seen more than one promising project vanish into the ether, simply because they failed to realize the importance of social media and promotion.

The Story Of An Emmy-Winning Duo On How To Bring Pro IP Creation To NFTs

In 1977’s Star Wars: A New Hope, Princess Leia appeared hovering over the table as a semitransparent volumetric image projected by a standalone robot R2D2, thus introducing a prevalent yet unrealistic notion of “holograms” and “holo calls” into mass culture. This is an example of how a fictional technology becomes an iconic representation of what future communication looks like. Despite this, a number of companies are now working on bringing a new dimension into our communications, including holographic video calls. But before we dive into the many marvelous opportunities the future is about to bring, let’s briefly recall what holography is all about (it’s not as close to the Star Wars-style holograms as you might think!). What are holography, holograms, and video holograms? (in its primary meaning) is a static photographic recording of a light field, formed by a laser split into two beams. Unlike a flat image or a photograph, you can view a hologram from different angles – by changing its position in your hands or looking at it from different viewpoints if it’s hanging on the wall, for example. Today you can see a lot of 3D images with holographic effects (i.e. on credit cards, postcards, etc). A hologram Two photos of a hologram from different viewpoints Developed in 1947 by Dennis Gabor, holography (from the Greek word “holos,” meaning “complete”) is a technology enabling 3D capturing and recreating an optical copy of a static object by means of laser interference. So, the moving holograms, like the ones featured in the Star Wars franchise – although inspired by existing technology – are more of a fantastic nature (alas!). The process of recording a hologram But what do we have when it comes to real-life solutions in holography? The project, for instance, has created holographic displays capable of demonstrating stereoscopic images by rendering different views according to a viewer’s position. Looking Glass Factory An alternative method of  “multiplex”, or , was developed by Lloyd Cross, who created moving 3D images from movie reels. The technology is based on creating an optical illusion by projecting a video image of an animated 3D model onto a semitransparent mirror or a screen set at a certain angle – so the image is actually flat but appears as three-dimensional. To create the 3D effect and 'the depth', the projected video content has a black background, which creates the illusion of three-dimensionality, the parallax effect. video holograms Although AR technologies also deal with 3D models “added” to the real world, such objects are not usually considered proper holograms since viewing them requires additional devices like AR glasses or smartphones which mix the real world and virtual environment data. For example, 8i such “holograms” by capturing the client’s movements with a bunch of cameras and rendering a moving 3D model, or an asset, which can then be integrated into various environments. creates Star Wars-style Holo Calls Clearly, it’s not possible to recreate the Star Wars-like holo calls by means of real-life holography – but still, this is a good example of a fantastic user experience. Imagine calling a friend or a family member and seeing them appear on your table like in the movie – as a semi-transparent image projected by an external light source! But, however impressive this kind of video call might seem. Over the last 44 years, since they were first demonstrated in Star Wars, they never actually emerged as real-life technology. This is especially upsetting in times of social distancing when everyone is fed up with standard Zoom calls on plain LCD screens. This problem was even mocked in an episode of the – a popular comedy TV series about technological startups: Silicon Valley Google has been a “video-booth” technology that enables the so-called volumetric video calls with external backlighting slightly resembling the Star Wars style. However, it still requires bulky and expensive equipment, a large specialized screen, several cameras for volumetric shooting, sensors, and the booth itself. Besides, the 3D illusion fades away when moving to the side just a few inches. developing Holo video calls in the Google secret lab Recognizable backlighting on the wall panels in Star Wars Given its price and size, it doesn’t seem a suitable solution for mass distribution of a technology enabling Star Wars-like holo calls with your friends... In 2018, Voxon Photonics, in cahoots with Verizon, made an experiment which they claim to be . Using the 5G network and the Intel RealSense depth camera, they carried out a real-time video conference where the caller’s volumetric face appeared on a 3D volumetric display. Well, that’s rather impressive, but still, this is definitely not intended for mass use. The screen is too big while the image is too small and is lacking the airy transparency of a Star Wars hologram. the first international truly holographic call Holographic Entertainment Music shows and DJ sets with laser lighting displays have been immensely popular since they were first introduced by such performers as Pink Floyd and Didier Marouani back in the 1970s. 3D images and visualizations are being projected on a smoked space or, more recently, on a transparent projection screen or a large transparent LCD screen installed between the performer and the audience. FMBooking in cooperation with Iboga Records has been making hologram shows for electronic dance music fans, adding mesmerizing visual effects to the journey that music provides. The shows feature an abundance of dynamic visual forms, colors, and textures that can be perceived from . all viewing angles Another example of Holo entertainment technology is life-sized hologram performances developed by . A medium-sized screen with a built-in projector is available for venues or corporate events along with a catalog of pre-recorded video holograms of DJs. Holocryptics Eric Prydz’s Holo Show has gained enormous popularity, not least because of its impressive visuals. But there are two problems with the show. First, the spectators are so eager to record the ‘hovering’ Star Wars-like music visualizations on the transparent screen that they stick to their smartphones and end up viewing the whole performance through its tiny screen. Second, usually the tickets are out in 10 minutes! already sold Would you like to have your personal Holo Show (i.e. the real-time art visualization of music rendered from any source on a transparent screen) – like Eric Prydz’s – but right in your home or backyard, where you are always in the front row, can watch shows at any time, and play any of your favorite music. The emergence of mass-market holo screens can reveal fascinating opportunities for music and art fans around the world. The spectacular holo shows in your own house or in the backyard, modern audio-visual installations, special TV shows as well as live performances being broadcast to home devices will now become available for everyone in each corner of the Earth — with compact holo screens and multimedia projectors. This sounds especially promising in the age of social distancing. A Holo Experience That's Accessible for Everyone is Coming Soon Before television emerged, people could watch movies, cartoons, documentaries or newscasts only in the cinema. When TVs were first introduced on the mass market, they had tiny screens – and yet, they were seen as a kind of miracle by families and friends and so they would gather together for a new thrilling experience. Today, we’re at the edge of a new era in home entertainment – with the emergence of portative holo screens tailored for holographic content. If the idea of personal holo entertainment seems too fantastic to you, here’s the good news — an affordable holo experience, including Star Wars-like holo calls, holo shows synced to your favorite music and innovative way to enjoy live streams, will soon be available with the launch of a brand-new projection display by (Hologram + Collider). The project is right now. Hollider running a campaign on KickStarter This transparent, frameless and “hovering” holo screen challenges gravity thanks to its impressive console design. At the same time, it exactly nails the need for the widely affordable means of making sci-fi-style video calls. You’ll only need a simple pico or mini multimedia projector widely available on the market and the Hollider screen for an utterly fascinating user experience. The adjustable construction of the transparent rear projection screen is designed to facilitate and simplify the assembling and operating of the holographic screen for a user. make holo-calls with your friends or colleagues, enjoy a personal Hologram Show synched to your favorite music in real time, watch special HOLO TV shows or just ordinary TV shows in the futuristic style on a holo screen, create a sci-fi interior design featuring 3D animated sculptures and collages at home or in the office, watch the world’s first audio-visual show streaming live to home holo screens. Some may argue that “this screen is too simple.” But, microchips were simple, too, and so was the pasteurization process – however, they both changed the world. The point here is a new vision and initiative rather than technical sophistication. The future is coming!

Are Star Wars-like Holo Calls Coming in the Near Future?

In the on-demand market, ride-hailing is one of the key services that account for a higher percentage of total revenue. Uber is the first ride-hailing app that entered the market and following that many taxi services launched their business in the market. Many taxi apps became successful and are still sustaining in the market. There is so much competition in the taxi service industry. Many small scale taxi services are competing with each other to become the key players in the market. Followed by the ride-hailing service, the robot-taxi service set its foot in the on-demand market. is the first and the only company to date that has launched the robot-taxi service. The main concept behind this business is to provide completely self-driving car services to the customers. Many people welcomed this idea, but Waymo’s service was limited to Pheonix, Arizona. It was launched on December 5, 2018. Waymo It has been one year since the company was founded, and it has completed over 100,000 rides to date. Currently, more than 1500 customers are using the service in a month. The cars used for this service are built based on sensor technology. These cars were being developed in Google's secret lab since the year 2009. It started testing a few years back and drove millions of miles to check if the self-driving feature is working properly. It was mostly tested on the roads of US cities. Why self-driven cars? The main motive in building cars with self-driving technology is to avoid accidents and for the safety of the people. A few stats about road accidents in the US are given below: More than 1.2 million people lose their lives in the roadways due to major accidents. Collisions that occur during traffic kills more than 35,000 people in a year.It is also found that more than 94% of these accidents occur because of human errors. So, Waymo was launched with safety of the people as the prime agenda. When it comes to self-driven cars, there will be no human intervention, and thereby, the accidents will lower drastically. What are the cars that Waymo is testing? Waymo has built Lexus Suvs and other custom-built prototypes. Recently, it has built Chrysler Pacifica minivan. It collaborated with Fiat company in their facility in Michigan. It tested all its prototypes on the roads of California, Texas, Washington, and Arizona with a test driver in each of the cars. They have completed over 200 hours of test driving in extreme weather conditions and rough roads. Right now, the cars of Waymo has around 300 years of driving experience. What is the technology employed in self-driven cars? The self-driven cars built by Waymo have a set of sensors and software incorporated in it. It also has a custom computer and telematics. These were tested with the SUVs before they were employed in the minivan for testing. Its sensors detect objects and vehicles in a few seconds. It looks out on all the four sides of a car and then drives accordingly. For instance, if a cyclist or a pedestrian is going to cross the road, then he/she will wave their hand before they cross. The sensors employed in the car will detect the motion of the hand and also predict the direction in which the cyclist or the pedestrian will move. Based on this analysis, the car will plan its next step. It can also scan the colour of the signal that is displayed. It stops when the colour is red, gets ready to move forward when the colour is yellow, and then starts driving once the colour changes to green. It plans each of its moves only after completely analyzing everything around the car. So, it is highly safe to travel in these self-driving cars. It will navigate safely in traffic and you do not have to put any effort into driving. What are their future plans? Their latest advancement is that they have launched an app for iOS device users. The app was previously available only on the Play Store. To mark its one year anniversary, Waymo starting developing a ride-hailing app and released the iOS app in the App Store. Upcoming entrepreneurs can by understating the demands of market. launch a ride-hailing app like Waymo Currently, they are planning to launch an Early Rider Program as a beta tester. People can enroll in this program and sign nondisclosure agreements with Waymo. They will then be put on the waitlist, and after a few days, they will be provided full access to driverless rides. They are planning to expand their service to all the cities in the US. They also have plans to cater to the needs of the trucking and delivery services. Their main goal is to avoid accidents and provide a safe journey to the customers. In the near future, they want to extend this service globally. Once the testing process is completely over, it will be available for use by the people. After a successful launch, it will be the first robot-ride hailing taxi app in the world.

Insights From Waymo: Self-Driving Taxis Are The Future of Ride-Hailing

How the Rad Lab at MIT was more important than Los Alamos in defeating the Nazis; Radar Won the War, the Atom Bomb Ended It Most of us have heard the story of the Manhattan Project, which was the gargantuan effort to take a scientific theory (a sustained nuclear chain reaction) and turn it into a reality (the atomic bomb) for the war effort.  In one of the largest scientific mobilizations ever, a number of physicists and other scientists were shipped off to Los Alamos, New Mexico in one of the most secret labs in the nation. The story has been told in many books and movies; my personal favorite is , an account that stretches from Leo Szilard’s initial idea, to Oppenheimer and General Groves setting up the lab on an old boys ranch in New Mexico. Day One That secret lab eventually built both a uranium and a plutonium bomb (nicknamed “Fat Man” and “Little Boy”).  The effort actually went well beyond the Los Alamos laboratory – for one thing they needed enough purified materials (Uranium 235 and Plutonium, both of which had to be separated); this involved the efforts of labs in Oak Ridge Tennessee, Berkeley (under the direction of Nobel prize winner Ernest Lawrence whose cyclotrons played an important role) and dozens of other locations. The reason many of these scientists participated was because they were afraid the Nazis would develop the bomb first. Though every schoolboy knows that the atomic bomb helped end the war, it seems like the Japanese may have been on their last legs anyways and Germany had already surrendered by this time. The real impact of the Manhattan project and nuclear power came after the war; in the Cold War and the nuclear arms race (which involved more destructive Hydrogen bombs). On the other hand, most schoolchildren (including most MIT students) don’t know the full story of the Rad Lab, another laboratory that had an equal, if not greater, impact on the outcome of World War II. This was the “radiology lab” at MIT which was housed in a “temporary building”, called building 20. In this lab, many of the nation's leading scientists perfected radar, building it from a crude tool of bouncing very large (and imprecise) radio waves into the highly accurate system for tracking airborne and seaborne objects (using centimeter length waves) which we think of today as a staple of all air and sea travel. Radar stands for , and while its history goes back to the nineteenth century when Hertz bounced radio waves off of metallic substances, it was the efforts of the Rad Lab scientists that turned it into the must-have, reliable tool it is for today's military and civilians. RAdio Detection And Ranging I actually got to know this building well - I spent many late nights there working on my thesis when I was a student at MIT; While I knew that the ratty old building had something to do with radar and world war two, I didn’t know the full history until recently. It turns out that the Rad Lab was a continuation of the work of a privately owned predecessor lab, known as the Loomis Laboratory. In her excellent book, nnet Conant, writes of this private laboratory that was set up Alfred Loomis, a wall street banker who became one of the wealthiest businessmen in the country (and later, a respected scientist in his own right). Tuxedo Park, Je Loomis started funding scientific projects at his private lab and made many discoveries related to radio waves and microwaves, including the first EEGs, with the help of professional scientists like Robert Wood (From Johns Hopkins), George Kistiakowsky (who would play an important role in both radar and nuclear weapons development), and Ernest Lawrence, whose cyclotron Loomis helped fund and get off the ground. As Germany started to attack the British Isles  (well before the US had entered the war), Loomis and others saw the importance of helping England hold off the German attacks using innovations in science. The German U-boats (submarines) were sinking British American ships in the Atlantic, and the German bombers had shifted from daytime to nighttime bombing of London, making it very difficult for the Brits to defend themselves. The development of radar proceeded quickly during this time only because men like Loomis were pushing its development from a theory, incorporating some of the top findings from the British physics community, who sent a mission to Washington in 1940 to share information with "the yanks" since they were on the verge of being defeated by the Nazis. The US military, it turns out, was hesitant to use this untested technology (as they were initially hesitant about the atomic bomb, thinking it was a weapon for the “next war”). Loomis, Vannevar Bush (who was head of the Carnegie Institute), Karl Compton (President of MIT), and  many others were put into a committee called NDRC (National Defense Research Committee) by FDR.  This is the kind of bipartisanship we don't see much of today, since many of these men were not Democrats nor supporters of FDR. The NRDC was setup specifically to oversee scientific developments that could impact the war effort – and two of the main committees were the “radar committee” and the “uranium committee”.  Since WWI, which was the first mechanized war, forward thinking members of the scientific and political establishment had come to realize that science and technology could be the decisive factor in winning wars. Like many accomplishments in our nations history, much of the technology developed in WWII was from fear that others (namely, the Germans at thisz time), might get there first. According to , Loomis was instrumental not just in setting up this lab but in pushing its research forward, even before there was adequate funding from the government. He did this by using his own funds and cajoling his contacts in industry to develop prototypes when formal funding hadn’t been approved or before the military had even agreed to use the equipment. Tuxedo Park If Loomis hadn’t done this, it’s very possible that the radar advancement would've gotten caught up in government bureaucracy , and radar development may not have happened fast enough to help England (and later the Americans) in the course of the war. In the book, Conant tells the story of Loomis, Lawrence, and many other prominent scientists and their impact on the development of this revolutionary technology in detail. While some primitive radar existed in both Britain and the US before the war, the real issues were in miniaturization  - both in getting the size of the waves down (for better accuracy)  and the size of the senders and receivers so they could become airborne and practical (originally radar was not very accurate and required separate sending and receiving devices, which were very large antennas). It turns out the British had one part of the answer, which was a magnetron, and the physicists that descended on Cambridge, MA at the Rad Lab en masse solved many of the practical problems of how to send and receive radio waves from the same device ,and came up with dozens of miniaturized models of this new technology . The government funding finally came through and this temporary building they had set up at MIT, building 20 (it never had a “real” name because it was meant to only be temporary) became a full-fledged operation churning out one advancement after the next. The lab kept expanding as Loomis and his team kept coming up with new applications – airborne radar to help bombers locate bombing targets, airborne radar for submarine detection, which allowed England to stop the bleeding of ships from the German submarine menace, and airborne bomber to bomber navigation which played a critical role in reducing the damage done to England by German’s formidable air force. Before airborne radar was a reality,  clunky ground-based radar was used to determine the approximate location of the German incoming bombers and then these coordinates were transmitted verbally to British pilots, which made it pretty difficult for them to intercept the German pilots. The innovations in the Rad Lab were critical to stopping the bombing of London and the UK by intercepting first German aircraft, and later in the war, the other deadly new weapon, the guided missile (or V-1 missiles that were shot by Germany at the UK). The temporary "building 20" at MIT  where the Rad Lab lived Loomis was able to attract some of the top physicists to the lab because of his association with men like Ernest Lawrence, who earned the Nobel prize because of the cycolotron Loomis helped him build. This lab wasn’t technically under the jurisdiction of the military (unlike the Los Alamos lab during the war), and as a result, it was a free-wheeling, skunkworks like operation where the scientsts were free to "invent" whatever they thought they might need. This was partly because the military didn’t realize it needed radar until after the US was dragged into the war and air power was finally recognized as the "new reality" of war.  If anything, the bombing of Pearl Harbor on December 7, 1941 showed the military just how vulnerable US forces were to air power. Although airplanes  had been present in WWI (the flying “aces”), it didn't become the decisive factor in warfare it's considered today until  the second World War. Without developments related to radar at the Rad Lab, the Allies would have  been at the mercy of the German war machine, which in many ways was ahead of the Allies at the beginning of the war. It is a wonder that the Nazis, who were so great at creating offensive weapons, didn’t get to have this advantage first.  Without the work of the Rad Lab,it’s very possible that the Nazis would have defeated Britain and taken over the UK before Pearl Harbor, which means we might be in a very different world today (If you want to see what this might be like, read Philip K Dick’s Man in the High Castle, or watch the Amazon TV series)! When the military finally started to take the uranium bomb idea more seriously (in the parallel Uranium committe of the NDRC), Loomis and Lawrence played key roles, helping to recruit many prominent physicists from the Rad Lab to the Manhattan project.  For example, one of Lawrence’s key proteges  from Berkeley, Luis Alverez, played a critical role in the development of many of the radar systems that went into production. This led  to a new billion-dollar industry when the United States entered the war with the Army and the Navy realizing the new reality and asking for radar systems on .  Alvarez himself would later go on to be part of the Manhattan project and receive the Nobel prize in physics. every single ship and every single airplane Similarly, another of Lawrence’s apprentices at Berkeley, Robert Oppenheimer (known as “Oppie” to scientists) went on to become the director of the Los Alamos lab which produced the atomic bomb. Since most of the basic research around radar had been done at this point, and the Rad Lab was so well funded, others were able to take over and finish production models after key physicists went to Los Alamos to help solve bomb-related problems. The lab was able to take the initial research and crank out all types of anti-aircraft, anti-submarine, anti-ship and anti-missile technology, not to mention the precursor to today’s GPS systems, all by using radio waves. After Germany was defeated and radar was in broad use by Allied forces, the decision was finally made to shut down the Rad Lab on August 5, 1945, since the technology was now in the hands of commercial manufacturers. This decision was followed with a number of press releases and announcements about the important accomplishments of this lab and the scientists who had been there in the war, but these were upstaged by the momentous news of the atomic bomb drop the next day. For example, according to Conant, Time Magazine was going to have a cover story about the work of the lab on radar and its decisive role in helping England stand up against the (seemingly) overwhelming numbers of German submarines and bombers. After the war, MIT chose not to tear down this temporary building right away as a testament to how important the Rad Lab had been to the war. When I was a student there in the late 1980s/early 1990s, I remember being surprised at how often, when I wanted to find out about some new lab or program, it would be housed in this shoddy old building. It turned out that it continued to be a “skunkworks” type building for MIT, housing new labs and departments for many years, ranging from Minsky’s work on linguistics, to the Experimental Study Group, to Doc Edgerton’s strobe lights.  Basically, something that didn’t have budget or office space elsewhere ended up in this “temporary” building.   It was finally torn down in 1998 for the new Stata center, which houses the computer science department at MIT. Conant writes near the end of Tuxedo Park: “The men who worked on the atomic bomb were hailed as heroes, and countless books and Hollywood movies would recount their exploits, while the daring and inventive minds who create radar were large forgotten… Only the Rad Lab veterans knew better, knew that if radar had not kept the Germans from defeating  England, the war might have been over before America entered the contest.  Everyone who had worked at the laboratory understood the decisive role their deadly devices. Had played in speeding the day of victory, and it was a reflected in a remark by DuBridge (later President of CalTech, who was director of the Rad Lab) that became something of an  unofficial slogan, their badge of honor: " Radar won the war; the atom bomb ended it.

The Other Secret Science Lab That Helped Win WWII

Is AI Secretly Reinforcing Bias and Inequality?

Authors:
(1) Rafael Rafailo, Stanford University and Equal contribution; more junior authors listed earlier;
(2) Archit Sharma, Stanford University and Equal contribution; more junior authors listed earlier;
(3) Eric Mitchel, Stanford University and Equal contribution; more junior authors listed earlier;
(4) Stefano Ermon, CZ Biohub;
(5) Christopher D. Manning, Stanford University;
(6) Chelsea Finn, Stanford University. Table of Links Abstract and 1. Introduction 2 Related Work 3 Preliminaries 4 Direct Preference Optimization 5 Theoretical Analysis of DPO 6 Experiments 7 Discussion, Acknowledgements, and References Author Contributions A Mathematical Derivations A.1 Deriving the Optimum of the KL-Constrained Reward Maximization Objective A.2 Deriving the DPO Objective Under the Bradley-Terry Model A.3 Deriving the DPO Objective Under the Plackett-Luce Model A.4 Deriving the Gradient of the DPO Objective and A.5 Proof of Lemma 1 and 2 A.6 Proof of Theorem 1 B DPO Implementation Details and Hyperparameters C Further Details on the Experimental Set-Up and C.1 IMDb Sentiment Experiment and Baseline Details C.2 GPT-4 prompts for computing summarization and dialogue win rates C.3 Unlikelihood baseline D Additional Empirical Results D.1 Performance of Best of N baseline for Various N and D.2 Sample Responses and GPT-4 Judgments D.3 Human study details Abstract While large-scale unsupervised language models (LMs) learn broad world knowledge and some reasoning skills, achieving precise control of their behavior is difficult due to the completely unsupervised nature of their training. Existing methods for gaining such steerability collect human labels of the relative quality of model generations and fine-tune the unsupervised LM to align with these preferences, often with reinforcement learning from human feedback (RLHF). However, RLHF is a complex and often unstable procedure, first fitting a reward model that reflects the human preferences, and then fine-tuning the large unsupervised LM using reinforcement learning to maximize this estimated reward without drifting too far from the original model. In this paper we introduce a new parameterization of the reward model in RLHF that enables extraction of the corresponding optimal policy in closed form, allowing us to solve the standard RLHF problem with only a simple classification loss. The resulting algorithm, which we call Direct Preference Optimization (DPO), is stable, performant, and computationally lightweight, eliminating the need for sampling from the LM during fine-tuning or performing significant hyperparameter tuning. Our experiments show that DPO can fine-tune LMs to align with human preferences as well as or better than existing methods. Notably, fine-tuning with DPO exceeds PPO-based RLHF in ability to control sentiment of generations, and matches or improves response quality in summarization and single-turn dialogue while being substantially simpler to implement and train. 1 Introduction Large unsupervised language models (LMs) trained on very large datasets acquire surprising capabilities [11, 7, 40, 8]. However, these models are trained on data generated by humans with a wide variety of goals, priorities, and skillsets. Some of these goals and skillsets may not be desirable to imitate; for example, while we may want our AI coding assistant to understand common programming mistakes in order to correct them, nevertheless, when generating code, we would like to bias our model toward the (potentially rare) high-quality coding ability present in its training data. Similarly, we might want our language model to be aware of a common misconception believed by 50% of people, but we certainly do not want the model to claim this misconception to be true in 50% of queries about it! In other words, selecting the model’s desired responses and behavior from its very wide knowledge and abilities is crucial to building AI systems that are safe, performant, and controllable [26]. While existing methods typically steer LMs to match human preferences using reinforcement learning (RL), we will show that the RL-based objective used by existing methods can be optimized exactly with a simple binary cross-entropy objective, greatly simplifying the preference learning pipeline. At a high level, existing methods instill the desired behaviors into a language model using curated sets of human preferences representing the types of behaviors that humans find safe and helpful. This preference learning stage occurs after an initial stage of large-scale unsupervised pre-training on a large text dataset. While the most straightforward approach to preference learning is supervised fine-tuning on human demonstrations of high quality responses, the most successful class of methods is reinforcement learning from human (or AI) feedback (RLHF/RLAIF; [12, 2]). RLHF methods fit a reward model to a dataset of human preferences and then use RL to optimize a language model policy to produce responses assigned high reward without drifting excessively far from the original model. While RLHF produces models with impressive conversational and coding abilities, the RLHF pipeline is considerably more complex than supervised learning, involving training multiple LMs and sampling from the LM policy in the loop of training, incurring significant computational costs. In this paper, we show how to directly optimize a language model to adhere to human preferences, without explicit reward modeling or reinforcement learning. We propose Direct Preference Optimization (DPO), an algorithm that implicitly optimizes the same objective as existing RLHF algorithms (reward maximization with a KL-divergence constraint) but is simple to implement and straightforward to train. Intuitively, the DPO update increases the relative log probability of preferred to dispreferred responses, but it incorporates a dynamic, per-example importance weight that prevents the model degeneration that we find occurs with a naive probability ratio objective. Like existing algorithms, DPO relies on a theoretical preference model (such as the Bradley-Terry model; [5]) that measures how well a given reward function aligns with empirical preference data. However, while existing methods use the preference model to define a preference loss to train a reward model and then train a policy that optimizes the learned reward model, DPO uses a change of variables to define the preference loss as a function of the policy directly. Given a dataset of human preferences over model responses, DPO can therefore optimize a policy using a simple binary cross entropy objective, producing the optimal policy to an implicit reward function fit to the preference data. Our main contribution is Direct Preference Optimization (DPO), a simple RL-free algorithm for training language models from preferences. Our experiments show that DPO is at least as effective as existing methods, including PPO-based RLHF, for learning from preferences in tasks such as sentiment modulation, summarization, and dialogue, using language models with up to 6B parameters. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. Authors: (1) Rafael Rafailo, Stanford University and Equal contribution; more junior authors listed earlier; (2) Archit Sharma, Stanford University and Equal contribution; more junior authors listed earlier; (3) Eric Mitchel, Stanford University and Equal contribution; more junior authors listed earlier; (4) Stefano Ermon, CZ Biohub; (5) Christopher D. Manning, Stanford University; (6) Chelsea Finn, Stanford University. Authors: Authors: (1) Rafael Rafailo, Stanford University and Equal contribution; more junior authors listed earlier; (2) Archit Sharma, Stanford University and Equal contribution; more junior authors listed earlier; (3) Eric Mitchel, Stanford University and Equal contribution; more junior authors listed earlier; (4) Stefano Ermon, CZ Biohub; (5) Christopher D. Manning, Stanford University; (6) Chelsea Finn, Stanford University. Table of Links Abstract and 1. Introduction Abstract and 1. Introduction 2 Related Work 2 Related Work 3 Preliminaries 3 Preliminaries 4 Direct Preference Optimization 4 Direct Preference Optimization 5 Theoretical Analysis of DPO 5 Theoretical Analysis of DPO 6 Experiments 6 Experiments 7 Discussion, Acknowledgements, and References 7 Discussion, Acknowledgements, and References Author Contributions Author Contributions A Mathematical Derivations A.1 Deriving the Optimum of the KL-Constrained Reward Maximization Objective A.1 Deriving the Optimum of the KL-Constrained Reward Maximization Objective A.2 Deriving the DPO Objective Under the Bradley-Terry Model A.2 Deriving the DPO Objective Under the Bradley-Terry Model A.3 Deriving the DPO Objective Under the Plackett-Luce Model A.3 Deriving the DPO Objective Under the Plackett-Luce Model A.4 Deriving the Gradient of the DPO Objective and A.5 Proof of Lemma 1 and 2 A.4 Deriving the Gradient of the DPO Objective and A.5 Proof of Lemma 1 and 2 A.6 Proof of Theorem 1 A.6 Proof of Theorem 1 B DPO Implementation Details and Hyperparameters B DPO Implementation Details and Hyperparameters C Further Details on the Experimental Set-Up and C.1 IMDb Sentiment Experiment and Baseline Details C Further Details on the Experimental Set-Up and C.1 IMDb Sentiment Experiment and Baseline Details C.2 GPT-4 prompts for computing summarization and dialogue win rates C.2 GPT-4 prompts for computing summarization and dialogue win rates C.3 Unlikelihood baseline C.3 Unlikelihood baseline D Additional Empirical Results D.1 Performance of Best of N baseline for Various N and D.2 Sample Responses and GPT-4 Judgments D.1 Performance of Best of N baseline for Various N and D.2 Sample Responses and GPT-4 Judgments D.3 Human study details D.3 Human study details Abstract While large-scale unsupervised language models (LMs) learn broad world knowledge and some reasoning skills, achieving precise control of their behavior is difficult due to the completely unsupervised nature of their training. Existing methods for gaining such steerability collect human labels of the relative quality of model generations and fine-tune the unsupervised LM to align with these preferences, often with reinforcement learning from human feedback (RLHF). However, RLHF is a complex and often unstable procedure, first fitting a reward model that reflects the human preferences, and then fine-tuning the large unsupervised LM using reinforcement learning to maximize this estimated reward without drifting too far from the original model. In this paper we introduce a new parameterization of the reward model in RLHF that enables extraction of the corresponding optimal policy in closed form, allowing us to solve the standard RLHF problem with only a simple classification loss. The resulting algorithm, which we call Direct Preference Optimization (DPO), is stable, performant, and computationally lightweight, eliminating the need for sampling from the LM during fine-tuning or performing significant hyperparameter tuning. Our experiments show that DPO can fine-tune LMs to align with human preferences as well as or better than existing methods. Notably, fine-tuning with DPO exceeds PPO-based RLHF in ability to control sentiment of generations, and matches or improves response quality in summarization and single-turn dialogue while being substantially simpler to implement and train. 1 Introduction Large unsupervised language models (LMs) trained on very large datasets acquire surprising capabilities [11, 7, 40, 8]. However, these models are trained on data generated by humans with a wide variety of goals, priorities, and skillsets. Some of these goals and skillsets may not be desirable to imitate; for example, while we may want our AI coding assistant to understand common programming mistakes in order to correct them, nevertheless, when generating code, we would like to bias our model toward the (potentially rare) high-quality coding ability present in its training data. Similarly, we might want our language model to be aware of a common misconception believed by 50% of people, but we certainly do not want the model to claim this misconception to be true in 50% of queries about it! In other words, selecting the model’s desired responses and behavior from its very wide knowledge and abilities is crucial to building AI systems that are safe, performant, and controllable [26]. While existing methods typically steer LMs to match human preferences using reinforcement learning (RL), we will show that the RL-based objective used by existing methods can be optimized exactly with a simple binary cross-entropy objective, greatly simplifying the preference learning pipeline. At a high level, existing methods instill the desired behaviors into a language model using curated sets of human preferences representing the types of behaviors that humans find safe and helpful. This preference learning stage occurs after an initial stage of large-scale unsupervised pre-training on a large text dataset. While the most straightforward approach to preference learning is supervised fine-tuning on human demonstrations of high quality responses, the most successful class of methods is reinforcement learning from human (or AI) feedback (RLHF/RLAIF; [12, 2]). RLHF methods fit a reward model to a dataset of human preferences and then use RL to optimize a language model policy to produce responses assigned high reward without drifting excessively far from the original model. While RLHF produces models with impressive conversational and coding abilities, the RLHF pipeline is considerably more complex than supervised learning, involving training multiple LMs and sampling from the LM policy in the loop of training, incurring significant computational costs. In this paper, we show how to directly optimize a language model to adhere to human preferences, without explicit reward modeling or reinforcement learning. We propose Direct Preference Optimization (DPO), an algorithm that implicitly optimizes the same objective as existing RLHF algorithms (reward maximization with a KL-divergence constraint) but is simple to implement and straightforward to train. Intuitively, the DPO update increases the relative log probability of preferred to dispreferred responses, but it incorporates a dynamic, per-example importance weight that prevents the model degeneration that we find occurs with a naive probability ratio objective. Like existing algorithms, DPO relies on a theoretical preference model (such as the Bradley-Terry model; [5]) that measures how well a given reward function aligns with empirical preference data. However, while existing methods use the preference model to define a preference loss to train a reward model and then train a policy that optimizes the learned reward model, DPO uses a change of variables to define the preference loss as a function of the policy directly. Given a dataset of human preferences over model responses, DPO can therefore optimize a policy using a simple binary cross entropy objective, producing the optimal policy to an implicit reward function fit to the preference data. Our main contribution is Direct Preference Optimization (DPO), a simple RL-free algorithm for training language models from preferences. Our experiments show that DPO is at least as effective as existing methods, including PPO-based RLHF, for learning from preferences in tasks such as sentiment modulation, summarization, and dialogue, using language models with up to 6B parameters. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. available on arxiv

Direct Preference Optimization: Your Language Model is Secretly a Reward Model

সরাসরি পছন্দ অপ্টিমাইজেশান: আপনার ভাষা মডেল গোপনে একটি পুরস্কার মডেল

Direkte Präferenzoptimierung: Ihr Sprachmodell ist insgeheim ein Belohnungsmodell

Optimización de preferencias directas: su modelo de lenguaje es en secreto un modelo de recompensa

Optimisation des préférences directes : votre modèle linguistique est secrètement un modèle de récompense

प्रत्यक्ष वरीयता अनुकूलन: आपका भाषा मॉडल गुप्त रूप से एक पुरस्कार मॉडल है

直接的な嗜好最適化: 言語モデルは実は報酬モデルである

직접 선호도 최적화: 언어 모델은 비밀리에 보상 모델입니다.

Otimização de Preferência Direta: Seu Modelo de Linguagem é Secretamente um Modelo de Recompensa

Прямая оптимизация предпочтений: ваша языковая модель — это тайная модель вознаграждения

Doğrudan Tercih Optimizasyonu: Dil Modeliniz Gizlice Bir Ödül Modelidir

Tối ưu hóa sở thích trực tiếp: Mô hình ngôn ngữ của bạn thực chất là một mô hình phần thưởng

直接偏好优化：你的语言模型其实是一个奖励模型

Reflections on the Decline of Science in England, and on Some of Its Causes by Charles Babbage, is part of the HackerNoon Books Series. You can jump to any chapter in this book . Section 3 here SECTION 3. OF THE SECRETARISHIPS. Whether the present Secretaries are the best qualified to aid in reforming the Society, is a question I shall not discuss. With regard to the senior Secretary, the time of his holding office is perhaps more unfortunate than the circumstance. If I might be permitted to allude for a moment to his personal character, I should say that the mild excellencies of his heart have prevented the Royal Society from deriving the whole of that advantage from his varied knowledge and liberal sentiments which some might perhaps have anticipated; and many will agree with me in regretting that his judgment has not directed a larger portion of the past deeds of the Councils of the Royal Society. Of the junior Secretary I shall only observe, that whilst I admit his industry, his perseverance, and his talents, I regret to see such valuable qualities exerted at a disadvantage, and that I sincerely wish them all the success they merit in situations more adapted for their developement. There are, however, some general principles which it may be important to investigate, which relate to the future as well as to the past state of the office of Secretary of the Royal Society. Inconvenience has already arisen from having had at a former period one of our Secretaries the conductor of a scientific journal; and this is one of the points in which I can agree with those who now manage the affairs of the Society. [These observations were written previous to the late appointment, to which I now devote Section 6. Experience seems to be lost on the Council of the Royal Society.] Perhaps it might be advantageous to extend the same understanding to the other officers of the Society at least, if not to the members of its Council. Another circumstance worthy of the attention of the Society is, to consider whether it is desirable, except in special cases, to have military persons appointed to any of its offices. There are several peculiarities in the military character, which, though they do not absolutely unfit their possessors for the individual prosecution of science, may in some degree disqualify such persons from holding offices in scientific institutions. The habits both of obedience and command, which are essential in military life, are little fitted for that perfect freedom which should reign in the councils of science. If a military chief commit an oversight or an error, it is necessary, in order to retain the confidence of those he commands, to conceal or mask it as much as possible. If an experimentalist make a mistake, his only course to win the confidence of his fellow-labourers in science, and to render his future observations of any use, is to acknowledge it in the most full and explicit manner. The very qualifications which contribute to the professional excellence of the soldier, constitute his defects when he enters the paths of science; and it is only in those rare cases where the force of genius is able to control and surmount these habits, that his admission to the offices of science can be attended with any advantage to it. Another objection deserving notice, although not applying exclusively to the military profession, is, that persons not imbued with the feelings of men of science, when they have published their observations, are too apt to view every criticism upon them as a personal question, and to consider that it is as offensive to doubt the accuracy of their observations as it is to doubt their word. Nothing can be more injurious to science than that such an opinion should be tolerated. The most unreserved criticism is necessary for truth; and those suspicions respecting his own accuracy, which every philosophical experimenter will entertain concerning his own researches, ought never to be considered as a reproach, when they are kept in view in examining the experiments of others. The minute circumstances and apparently trivial causes which lend their influence towards error, even in persons of the most candid judgment, are amongst the most curious phenomena of the human mind. The importance of affording every aid to enable others to try the merits of observations, has been so well expressed by Mayer, that I shall conclude these remarks with an extract from the Preface to his Observations: "Officii enim cujusque observatoris ease reor, de habitu instrumenti sui, de cura ac precautione, qua usus est, ad illud recte tractandum, deque mediis in errores ejus inquirendi rationem reddere publice, ut aliis quoque copia sit judicandi, quanta fides habenda conclusionibus ex nostris observationibus deductis aut deducendis. Hoc cum minus fecissent precedentis saeculi astronomi, praxin nimis secure, nimisque theoretice tractantes, factum inde potissimum est, ut illorum observationes tot vigiliis tantoque labore comparatae tam cito obsoleverint." P. viii. There are certain duties which the Royal Society owes to its own character as well as to the public, which, having been on some occasions apparently neglected, it may be here the proper place to mention, since it is reasonable to suppose that attention to them is within the province of its Secretaries. The first to which I shall allude is the singular circumstances attending the fact of the Royal Society having printed a volume of Astronomical Observations which were made at the Observatory of Paramatta (New South Wales), bearing the title of "The Third Part of the Philosophical Transactions for the Year 1829." Now this Observatory was founded at the private expense of a British officer; the instruments were paid for out of his purse; two observers were brought from Europe, to be employed in making use of those instruments, at salaries defrayed by him. A considerable portion of the observations so printed were made by these astronomers during their employment in his service, and some of them are personally his own. Yet has the Royal Society, in adopting them as part of its Transactions, omitted all mention, either in their title-page, preface, or in any part of the volume, of the FACT that the world owed these valuable observations to the enlightened munificence of Lieutenant-General Sir Thomas Brisbane; whose ardent zeal in the pursuit of science induced him to found, at his own private expense, an establishment which it has been creditable to the British Government to continue as a national institution. Had any kindred feelings existed in the Council, instead of endeavouring to shift the responsibility, they would have hastened to rectify an omission, less unjust to the individual than it was injurious to English science. Another topic, which concerns most vitally the character and integrity of the Royal Society, I hardly know how to approach. It has been publicly stated that confidence cannot be placed in the written minutes of the Society; and an instance has been adduced, in which an entry has been asserted to have been made, which could not have been the true statement of what actually passed at the Council. The facts on which the specific instance rests are not difficult to verify by members of the Royal Society. I have examined them, and shall state them before I enter on the reasoning which may be founded upon them. In the minutes of the Council, 26th November, 1829, we find— "Resolved, that the following gentlemen be recommended to be put upon the Council for the ensuing year." [Here follows a list of persons, amongst whom the name of Sir John Franklin occurs [Sir John Franklin was absent from London, and altogether unacquainted with this transaction, until he saw it stated in the newspapers some months after it had taken place. That his name was the one substituted for that of Captain Beaufort I know, from other evidence which need not be produced here, as the omission of the latter name is the charge that has been made.], and that of Captain Beaufort is not found. [Any gentleman may satisfy himself that this is not a mistake of the Assistant Secretary's, in copying, by consulting the rough minutes of that meeting of the Council, which it might perhaps be as well to write in a rough minute-book, instead of upon loose sheets of paper; nor can it be attributed to any error arising from accidentally mislaying the real minutes, for in that case the error would have been rectified immediately it was detected; and this has remained uncorrected, although publicly spoken of for months. As there is no erasure in the list, one is reluctantly compelled to conjecture that the real minutes of that meeting have been destroyed.]] Now this could not be the list actually recommended by the Council on the morning of the 26th of November, because the President himself, on the evening of that day, informed Capt. Beaufort that he was placed on the house list; and that officer, with the characteristic openness of his profession, wrote on the next or the following day to the President, declining that situation, and stating his reasons for the step. Upon the fact, therefore, of the suppression of part of a resolution of the Council, on the 26th of November, there can be no doubt; but in order to understand the whole nature of the transaction, other information is necessary. It has been the wish of many members of the Society, that the President should not absolutely name his own Council, but that the subject should be discussed fairly at the meeting previous to the Anniversary—this has always been opposed by Mr. Gilbert, and those who support him. Now, it has been stated, that, at the meeting of the Council on the 26th of November, the President took out of his pocket a bit of paper, from which he read the names of several persons as fit to be on the Council for the ensuing year;—that it was not understood that any motion was made, and it is certain that none was seconded, nor was any ballot taken on such an important question; and it was a matter of considerable surprise to some of those present, to discover afterwards that it was entered on the minutes as a resolution. This statement I have endeavoured to verify, and I believe it to be substantially correct; if it was a resolution, it was dictated, not discussed. It is also important to observe, that no similar resolution stands on the council-books for any previous year. On examining the minutes of the succeeding Council, no notice of the letter of Captain Beaufort to the President is found. Why was it omitted? If the first entry had been truly made, there would have been no necessity for the omission; and after the insertion of that letter, a resolution would naturally have followed, recommending another name instead of the one withdrawn. Such was the natural and open course; but this would have exposed to the Society the weakness of those who manage it. If the rough minutes of each meeting of the Council were read over before it separated, and were copied previously to the next meeting, such a substitution could hardly have occurred; but, unfortunately, this is not the case, and the delay is in some cases considerable. Thus, the minutes of the three Councils, held on February 4, on February 11, and on March 11, were not entered on the minute-books of the Council on Tuesday, the 16th March; nor was this the fault of the Assistant-secretary, for up to that day the rough minutes of no one of those Councils had been transmitted to him. Deeply as every friend to the Royal Society must regret such an occurrence, one slight advantage may accrue. Should that resolution be ever quoted hereafter to prove that the Council of 1829 really discussed the persons to be recommended as their successors, the detection of this suppression of one portion of it, will furnish better means of estimating the confidence due to the whole. About HackerNoon Book Series: We bring you the most important technical, scientific, and insightful public domain books. This book is part of the public domain. Charles Babbage (1998).Reflections on the Decline of Science in England, and on Some of Its Causes. Urbana, Illinois: Project Gutenberg. Retrieved https://www.gutenberg.org/cache/epub/1216/pg1216-images.html This eBook is for the use of anyone anywhere at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this eBook or online at , located at www.gutenberg.org https://www.gutenberg.org/policy/license.html

OF THE SECRETARISHIPS.

The Duty of American Women to Their Country by Catharine Esther Beecher, is part of the HackerNoon Books Series. You can jump to any chapter in this book . From the Hon. Thomas Burrowes, late Secretary of State in Pennsylvania. here FROM THE HON. THOMAS BURROWES, LATE SECRETARY OF STATE IN PENNSYLVANIA. I have long been of opinion that the of our age and country, in reference to the sound instruction of the coming generation, is the . great deficiency want of teachers I am now fully convinced that this want supplied any other step can be safely or usefully taken. Nay, I believe that, until this indispensable preliminary measure is accomplished, money, and effort, and legislation will be, , money, and effort, and legislation thrown away. Since 1834, this state has expended more than for the support of her common schools, and, at the end of ten years, I see but little improvement. must be before as they have been nearly five millions In this immense expenditure, not a dollar has been spent to secure this great prerequisite— ; and hence the system has not only failed to obtain general favour, but is in danger of becoming more and more unacceptable the longer it is tried. It is sad to think that we have thus wasted of dollars, and of time, to say nothing of the labour expended and obloquy encountered, and must now re-commence from the foundation; but so it is. good teachers five millions ten years I know of no cause which so much needs a as this. Let not its friends shrink from the undertaking because they may not be able to operate in all, or even in many of the states. Let it be remembered that if a commencement is made in one state, and a report of results sent forth, it will serve to start the good work in all the rest. general movement The necessities, the crying necessities of this cause, are far and away before those of the Temperance Reform, or of Colleges, or of Foreign Missions. He who, being fit, should devote himself to this cause, would confer a greater benefit on his fellow-man than he could possibly do by any other use of his time and talents. The missionary to a heathen land opens to his fellow-man; the missionary in this cause opens of his fellow-citizens, not only to the Book of Life, but to a knowledge of all those rights and duties, without which our free institutions cannot stand to encourage and reform the world. the Book of Life the mind If my gifts and domestic relations permitted, I should devote myself to a mission in this and other states for the purpose of impressing on Legislatures, philanthropists, and teachers, the . necessity of Teachers’ Seminaries A gentleman, supported to operate in this cause, might be employed in this way. He could visit different states one after another, and address the citizens of each county in the county town, after long and full notice. Besides addressing the people publicly, he could appeal to leading individuals privately, and engage them to act with him for this object. Meantime, he could be obtaining educational statistics for future use, and ere long he could make such a report as would set the people to work in earnest, and for their own sakes. While thus proceeding, he could also obtain the promise of one or more intelligent persons in each county, to write on the subject every week in each of the county newspapers. Articles thus addressed to the reason, the patriotism, and the of the people, would have a powerful effect, and cost nothing. economy If funds could be provided from private benevolence to establish proper in two or three states, they would set the matter far ahead in a few years. They would serve as and to the public, and would not long continue to need the support of private philanthropy. They would really be , or establishments. Teachers’ Institutions models inducements normal pattern Beyond a doubt, the plan ought to embrace institutions for the preparation of teachers. The gentleness, self-devotion, and untiring humanity of women eminently qualify them to be the instructers of the more youthful pupils of both sexes, and of their own of all ages. There is not a show of any reason why male teachers only should be provided for at the public charge, when female teachers are as necessary, as useful, and as much confided in by the public. female About HackerNoon Book Series: We bring you the most important technical, scientific, and insightful public domain books. This book is part of the public domain. Catharine Esther Beecher (2016). The Duty of American Women to Their Country. Urbana, Illinois: Project Gutenberg. Retrieved https://www.gutenberg.org/cache/epub/53739/pg53739-images.html This eBook is for the use of anyone anywhere at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this eBook or online at , located at . www.gutenberg.org https://www.gutenberg.org/policy/license.html

From the Hon. Thomas Burrowes, late Secretary of State in Pennsylvania.

The $30,000 Bequest, and Other Stories by Mark Twain, is part of the HackerNoon Books Series. You can jump to any chapter in this book . A LETTER TO THE SECRETARY OF THE TREASURY here A LETTER TO THE SECRETARY OF THE TREASURY Riverdale-on-the-Hudson, OCTOBER 15, 1902. The Hon. The Secretary Of The Treasury, : Washington, D. C. Sir,—Prices for the customary kinds of winter fuel having reached an altitude which puts them out of the reach of literary persons in straitened circumstances, I desire to place with you the following order: Forty-five tons best old dry government bonds, suitable for furnace, gold 7 per cents., 1864, preferred. Twelve tons early greenbacks, range size, suitable for cooking. Eight barrels seasoned 25 and 50 cent postal currency, vintage of 1866, eligible for kindlings. Please deliver with all convenient despatch at my house in Riverdale at lowest rates for spot cash, and send bill to Your obliged servant, Mark Twain, Who will be very grateful, and will vote right. About HackerNoon Book Series: We bring you the most important technical, scientific, and insightful public domain books. This book is part of the public domain. Mark Twain (2004). The $30,000 Bequest, and Other Stories. Urbana, Illinois: Project Gutenberg. Retrieved October 2022 https://www.gutenberg.org/cache/epub/142/pg142-images.html This eBook is for the use of anyone anywhere at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this eBook or online at , located at . www.gutenberg.org https://www.gutenberg.org/policy/license.html

A LETTER TO THE SECRETARY OF THE TREASURY

Recently, I read a that revealed a concerning trend regarding the political biases of . When analyzing various models, OpenAI's ChatGPT and leaned towards left-wing libertarianism, while Meta's LLaMA was identified as the most right-wing authoritarian. research study Large language models GPT-4 These findings suggest LLM models exhibit political biases regardless of the specific variant. Yet, in the face of such revelations, OpenAI responded differently than one might expect. However, when OpenAI faced criticism for potential liberal bias, they emphasized a neutral approach, calling any emergent biases "bugs, not features”. Additionally, asserts no language model can be free from political biases. Moreover, as researchers delve into the mechanisms behind AI language model bias, they’ve uncovered the actual reasons behind these biases in the models. PhD Researcher Chan Park LLM models can acquire biases through various mechanisms. Biased or unrepresentative training data can cause models to learn and amplify those biases. For example, if the data contains politically biased texts, the models may unintentionally adopt and reinforce those biases. Algorithmic design and optimization objectives also play a role in bias formation. The algorithms used in LLM models are designed to optimize certain objectives, such as language fluency or prediction accuracy. However, these objectives may inadvertently prioritize or favor certain biases in the training data, resulting in biased outputs. During fine-tuning, models are exposed to additional data specific to the task. If this data is biased or reflects certain political leanings, the models may further adopt and reinforce those biases. An interesting observation comes from a study comparing Google's BERT models with OpenAI's GPT models. The study revealed that BERT models exhibited more social conservatism, possibly due to training on conservative books. In contrast, GPT models leaned in a different direction due to training on liberal internet texts. This highlights the importance of careful data curation and algorithmic fairness in LLM model development. Considering this, the question arises: should AGI (Artificial General Intelligence) be required to publicize its training data? On the one hand, transparency is crucial in understanding and addressing biases. The ability to scrutinize the data and the training methods can lead to better accountability and potentially mitigate biases. It could provide a clearer picture of how AGI is shaped and help identify areas for improvement. However, there are valid concerns about privacy and intellectual property rights. Requiring complete disclosure of training data may hinder innovation and compromise sensitive information. At the heart of AI's relationship with bias and human nature lies a crucial point: While AI aims to be fair, it cannot escape the fact that it's and that data carries our biases. This poses a challenge that we must address. built on data created by people To tackle this challenge, we need transparency. It's important to openly share information about how AI models are trained, the data they use, and the decisions made during their development. This transparency helps us understand and address biases. However, we must be careful not to hinder progress or reveal sensitive information. Striking a balance between transparency and protecting important details is essential. Achieving this balance requires collaboration. Researchers, developers, policymakers, and society must work together. By having open discussions, following ethical practices, and promoting responsible AI development, we can ensure that innovation and accountability go hand in hand. "In the mirror of AI, we find not a reflection of perfection, but the echoes of our own biases."

Stories, News & Mentions about SECRETLAB

Secretlab’s Stories on HackerNoon

Secretlab's News & Mentions