Mediaboss Marketing

Image by Alex Anstey

Its a dang good thing were winning the cricket because the government has collapsed. Scott Morrisons Team Australia has left the health system to fail; the virus is out of control, tracking and testing has crashed, and Liberal Party corporate mates Harvey Norman and Chemist Warehouse are profiteering. Michael West looks at the price of Scomos personal responsibility.

They let it rip. And it ripped. Three Covid infections in Bali yesterday, 35,000 in NSW. Nursing homes are locking down so Australias elderly can once again die in peace, untroubled by the distraction of having their loved ones around them. Restaurants, a slew of businesses going belly up. Hospitals in such crisis that Covid-positive nurses are being called back to work.

Its ripped alright. Australia. Government, broken. Its a good thing there are things the government doesnt control because they are going well. The Ashes, the share market, iron ore prices. The stuff they do control however, but for which they are apparently not responsible, things like the health system, defence spending, nursing homes, quarantine; falling apart.

The single most important duty of government is to keep the people safe. Theyve failed.

They do control the media though, or most of it at least, and thats still going well for them. Somebody from the ABC was complaining that Scott Morrison, Greg Hunt, Dom Perrottet and a few of the personal responsibility crew knocked back an offer from 7.30 Report to go on tele to explain themselves. Naturally they refused.

Why would Scomo front Laura Tingle anyway when he can hop on the JobKeeper subsidised Seven Sunrise at do some public relations and marketing?

Day-in-day-out, he can rely on the corporate media at the government subsidised News Corporation, Sky News, Seven and Nine Entertainment to create the illusion everything is somebody elses fault.

It was only yesterday that Rupert Murdochs courtiers at The Australian were claiming free Covid tests are wrong. Nine media timidly accepted too that its big advertiser Harvey Norman, yet to pay back its JobKeeper despite record profits, was entitled to make a profit from selling Covid tests.

Then Morrisons black arts department in PMO chucked a 180 and suddenly it was PM to the rescue a bold red non-exclusive Exclusive about our ScoMo, genuflecting about the white knight riding to the rescue.

Chronic rapid test shortage to continue for days as PM offers states more free kits, splashed The Age and SMH about the PMs new plan.

Good thing they did not elect to do journalism and tell the truth; that the government is out of its depth, sinking the tumultuous seas of its own incompetence and crony capitalism, and the real plan is just to do a Donald Trump and keep the PR coming via its subsidised and compromised corporate media outlets.

Checking in on what our white knights up to, we strayed unwittingly this morning onto the website of the National COVID-19 Coordination Commission (NCCC).

The National COVID-19 Coordination Commission (NCCC) website and social media channels are now live, it declared. Alas, no they werent. Page not Found, was the search result. Quite symbolic really because what we are dealing with here is Government not Found.

Like most Coalition plans, the NCCC was secretive and didnt work out. Despite the grand ambitions and the curtseying media which accompanied its inception, the commission was quietly disbanded a few months later and after some lavish fees for the assorted fossil fuel directors and Liberal Party acolytes.

Thank heavens for independent media and independent politicians. This from Crikey: Last July the commission was granted a new mandate that allowed much of its work to be deemed cabinet in confidence meaning it could operate largely behind closed doors. This was challenged by independent Senator Rex Patrick.

The commission spent more than $1 million on market research, none of which has been released publicly. About $541,000 of that went to former Liberal Party pollster Jim Reed another $500,000 went to Boston Consulting Group (BCG) for management advisory services. BCG was also hired by the government to advise it on gas modelling. PricewaterhouseCoopers was also given $79,200 for management advisory services.

Plus cest la mme chose, as they say in France, where incidentally a new mutant strain of the virus has just bobbed up, reportedly originating in Cameroon. We dont know what this Ihu variant will bring but its a fair bet the government here will ignore the scientific and medical advice and the empirical evidence of how best to handle it.

It is also a fair bet that the looming election campaign will be more ludicrous than any before. There are countless examples but this weeks Murdoch oped claiming rapid test *should* be free, then the exclusive the very next day saying how great it was that the PM was providing free kits is a classic of the captured media genre.

This is not journalism, it is not media in any traditional sense, it is a foreign media company given to warmongering and running daily propaganda for the Coalition.

What the Coalition really needs is a good enemy. It is shaping as a Covid election otherwise, and even with the support of its media allies and a $16bn war-chest of publicly funded bribes to fork out to marginal seats, this will be a challenge because everybody now has a personal Covid experience and only the most strident of Tories will contend with any conviction that the government is competent.

It is fair to say that governing in this pandemic is no cakewalk. Mistakes will inevitably be made, and both public health and economic measures change according to changing circumstances. The problem for Australia is more systemic. This is a government which ignores sound, scientific advice on everything from climate and energy to health.

It is a government of business lobby marionettes, a government which listens to those who give the Coalition money. How else could it ignore medical advice on the Pandemic, or allow 50 new fossil fuel projects to be on the boil? How else could it blow up $40bn in JobKeeper hand-outs to Italian fashion houses and other foreign multinationals as well as large local corporations and other entities with record incomes.

So to the inevitable distractions. The China rhetoric will intensify. Comically, The Australian ran this during the week: Nuclear submarine pact has Beijing rattled. The governments outhouse PR operatives in the Murdoch media love a war and the China-threat stories are daily fodder. So much so that surveys find millions of Australians actually fear a Chinese invasion.

To address the actual facts though, there is no submarines deal. There is some sort of secret agreement with the US and Britain to buy submarine technology under the AUKUS treaty.

The claim that China is shaking in its boots thanks to Australias deal to get some subs sometime in the next 20 years will be a bit late for the Election in May but that wont stop the propagandists concocting an immediate threat and decisive action by the Morrison government.

If there was a deal we would probably know what sort of subs there were by now, US or UK technology, old or new. There would be a cost, which at this point is only an estimate of $90bn.

For its part, China has 60 or 70 submarines itself. It exports the things to at least four other nations. The very notion that Australia ought to go to war with China over Taiwan, or for that matter any other reason than an invasion of Australia which is a bizarre notion in itself is simply ridiculous. We would be shellacked in quick order, despatched like the English batting line-up, even with our entire GDP spent on weaponry.

If it were not so tragic it would be funny. Here is a media machine, arrant apologists for Donald Trump, which brought us the Wuhan lab theory, a Chinese conspiracy to spread the pandemic, then the Chinese military threat daily.

It can only be a matter of time before they join the dots between the shortage of RAT tests and Chinese profiteering. Meanwhile, there would be no profiteering if there was a competent government which reacted to independent, rational advice rather than whispers from the business lobby.

There is no public appetite for more lockdowns and the latest variant of the virus is less dangerous. Yet people will die as a result of the collapse of government and the failure of our elected leaders to take proper advice. Its not hard, just a matter of listening to people who know what they are talking about, seeing how other countries are managing public health, acting decisively and ignoring the siren calls of business lobbyists.

See the rest here:
They 'let her rip', and she ripped; government collapses in Australia - Michael West News

Former Centers for Disease Control and Prevention Director Robert Redfield denounced the Chicago Teachers Union's decision to return to virtual learning, arguing that the "safest place" for children right now is in the classroom.

The Chicago Teachers Union voted late Tuesday to return to full-time remote learning amid the surge in COVID-19 cases. Redfield told "America Reports" Wednesday that the decision has no scientific basis and only risks causing further harm to children grappling with the mental health impacts of the pandemic.

WHITE HOUSE REITERATES SUPPORT FOR OPEN SCHOOLS AFTER CHICAGO TEACHERS UNION VOTE

"Its so important to keepour schools open to face-to-facelearning. We can do it in a safe andresponsible way," Redfield emphasized. "The reality is the school isprobably the safest place forthese students to be, so I dontthink the decision really isgrounded in science. I dontthink its grounded in ourknowledge of what the situationis."

The Chicago Teachers Union's vote forced classes as early as Wednesday to be canceled. The vote was approved by 73% of the union's members, who voted for no in-class learning until cases of COVID-19 "substantially subside" or until union leaders approve an agreement for safety protocols with the district.

Redfield said the move will only exacerbate the negative effects children are experiencing, telling Fox News that there's "no question that the public healthinterest of K-12 students is notserved by remote learning."

Dr. Robert Redfield, director of the Centers for Disease Control and Prevention testifies at a hearing with the Senate Appropriations Subcommittee on Labor, Health and Human Services. Washington, Wednesday, Sept. 16, 2020. (Anna Moneymaker/New York Times, Pool via AP)

"Whether its nutritional supportthat millions of children get orthe mental health servicesupport that over 7 million kidsget, whether its the ability todetect child abuse, the mentalhealth, depression, loneliness,suicide, drug abuse," he continued. "Not tomention ... some of these kids fall off thelearning curve, and some of them are never going to get backon the learning curve."

"This is really not in theinterest of children," Redfield reiterated. "Publichealth interest is to keep thekids in face-to-face learning.It can be done safe andresponsibly. Its actually safer thanhaving them at home in thecommunity."

Redfield later addressed the newest guidance from the Centers for Disease Control and Prevention (CDC) on testing and isolation that has left many confused.

On Tuesday, the top health agency reiterated that children and adults who test positive can halve their isolation time from 10 to five days if they're asymptomatic. The CDC declined to add a clear testing recommendation while saying that people can take a test if they have "access" and "want to."

PSAKI INSISTS CDC GUIDED BY SCIENCE AMID SHIFTING COVID-19 GUIDANCE

"I agree with you,its highly confusing,"Redfield said, adding that while he is in favor of reducingthe isolation time period to five days,he is "totally not in agreementwith their decision not to do atest."

Centers for Disease Control and Prevention (CDC) Director Rochelle Walensky gives her opening statement during the Senate Health, Education, Labor and Pensions hearing. November 4, 2021. REUTERS/Elizabeth Frantz

"I think we really need toembrace a test they find, and ifyou are negative, you can test andreturn," he said. "If you are positive, you aregoing to need to get a testagain. I personally would not waituntil day 10, because the whole purposewas to get people back into theworkforce.You test at day five, and you arenegative, you go back to work.If you are positive, stay inisolation.Test at day seven and if youre now negative,you go back to work."

CLICK HERE TO GET THE FOX NEWS APP

Redfield said the updated guidance doesn't consider"knowledge of infection asfundamental to whether youreturn to work or the issue ofschools, [based on] what we call test and stay."

"Itscritical we use this testing asour guide," he said.

Here is the original post:
Former CDC Director Redfield: The safest place for children right now is in the classroom - Fox News

Parliament fire is finally under control

Parliament has confirmed the fire on the roof of the National Assembly is finally under control. The fire flared up on Monday afternoon. A devastating fire ripped through Parliament on Sunday.Parliament said in a statement:

It is with a great sense of relief that Parliament confirms the containment of the fire flare up at the roof of the National Assembly (NA) on Monday, and there has not been any further fire incident.

The last 24 hours had been critical, with firefighters closely monitoring and combing through the scene.

After the firefighters contained the fire at midnight yesterday, they remained on site, although at reduced capacity for monitoring the situation, to conduct a thorough assessment and to establish the extent of the damage caused.

Since Sunday, there have been 300 firefighters working shifts and over 60 fire engine vehicles. One fire engine remains at the scene currently, with five crew members working throughout all the floors, ensuring no flare-ups.

The firefighters will assess later this afternoon for possible total withdrawal from the site today and see if the building is safe to be handed over to the South African Police Services.

The Presiding Officers of Parliament, Ms Nosiviwe Mapisa-Nqakula and Mr Amos Masondo applauds the lionhearted firefighters. The firefighters fought to bring the fire at the Parliamentary precinct under control.

The extent of the damage in the NA is severe. The Presiding Officers confirm that efforts to save the Mace were successful yesterday after two days of the fire. It has been retrieved from its safe storage without any damage.

The Mace is an important symbol that signifies the authority and sitting of the NA. It is carried into the Chamber by the Serjeant-at-Arms and announces the arrival of the Speaker of the NA.

It signifies that the House is formally in session and that its proceedings are official. The Mace was designed to reflect the history, traditions, and diverse cultures and languages of South Africa. The design also celebrates its natural beauty, plant and animal life, and rich mineral resources.

The NA Speaker remains grateful for saving the Mace as its recreation could be difficult.

The Museum is also unharmed from the ravaging fire including artworks and heritage objects and the Keiskamma tapestry on the ground floor of the Old Assembly Building.

The Keiskamma Tapestry tells the South African story in beadwork, skins and embroidery from the perspective of ordinary people. It is 112 metres long and 70 metres high. Women from the Keiskamma Art Project, a community initiative and non-profit organisation in Hamburg, on the banks of the Keiskamma River in the Ngqushwa region of the Eastern Cape, made this artwork.

It is a powerful symbol of our people's Parliament. It demonstrates our support for women's empowerment and support for local initiatives. The former Speaker of the NA, Ms Baleka Mbete unveiled the tapestry in 2006 on International Women's Day, 8 March.

As part of the internal stakeholder meetings yesterday, the Presiding Officers briefed the leaders of political parties, the Chief Whips Forum and Nehawu leadership.

The Presiding Officers reassured everyone that no stone would be left unturned in getting to the bottom of how the incident happened. They said Parliament would conduct its internal investigation on any lapses that contributed to the incident.

"We don't want to make any speculations about what may have led to this incident, but we continue to be very concerned that the institution of Parliament and its symbols could be destroyed in this manner. So, we will demand full accountability and if so found, for those responsible to be punished", said the Presiding Officers.

They further asked all to work together despite the incident and find ways to ensure that Parliament's work continues and will not allow the disaster to derail the institution's important work.

Furthermore, they confirmed that the State of the Nation Address, Budget Speech, and other programmes would proceed as planned.

Parliament will share further details about where and how these events will occur.

See more here:
DEVELOPING | Firefighters withdraw from Parliament, buildings handed over to Hawks - News24

Surf Life Saving WA (SLSWA) is urging caution with medium risk ratings for this Saturdays second hour of the West Coast Zone (WCZ) abalone fishing season between 7am and 8am.

SLSWA modelling uses the best available information on the conditions from multiple sources and rates the risks for factors, such as wind speed, tide, swell height, and swell period.

The modelling recommends the fishing hour can go ahead on 8 January, with appropriate caution, as conditions are expected to appear favourable with a swell period of 11-12 seconds and waves expected to be breaking at near 1-1.5m height in most locations; particularly outside of the central metro area. However, the amount of water likely to be pushing across reef platforms remains a key concern.

Reef holes and drop offs can be hazardous for those with low swimming capabilities and fishers are also encouraged to wear appropriate clothing when collecting the abalone.

Moderate temperatures of 27-30 degrees are expected for this Saturdays fishing session, with an average wave height of around 1.5 metres, particularly in Peel and North Metro areas and moderate offshore winds, gusting to 18 knots.

Licensed abalone fishers who plan to take part in this Saturdays fishing hour in the WCZ between Moore River and the Busselton Jetty will still need to make their own evaluation of the sea and weather conditions on the day, to ensure they have the water skills to manage them.

Department of Primary Industries and Regional Development (DPIRD) Senior Management Officer Nick Blay said fishers should follow safety advice.

Personal safety should be the focus of abalone fishers at all times, Mr Blay said. SLSWA lifesavers will be monitoring the abalone fishing this Saturday, but each fisher must not take risks beyond their skill level in the water and on the reefs.

Apart from this Saturday, fishing between 7am and 8am, there will be two other abalone fishing sessions at the same time on Saturday 5 February and Saturday 19 February.

DPIRD Compliance officers will again be at WCZ abalone fishing locations this Saturday, to check fishers have the required licence and are complying with the rules.

More on abalone fishing rules is available at http://www.fish.wa.gov.au. SLSWA has abalone fishing safety tips at: https://www.mybeach.com.au/safety-rescue-services/coastal-recreation/abalone/.

We urge anyone heading to WAs beaches at any time to switch on their Sea Sense check http://www.sharksmart.com.au or download the SharkSmart WA app. The app combines latest shark activity, as well as beach safety features such as Surf Life Saving WA patrolled beaches and weather forecasts, to help people plan their trips to the ocean.

Media contact: Ashley Malone, DPIRD media liaison - mobile 0418 901 767

Go here to read the rest:
West Coast Zone abalone fishing season proceeds with caution - Government of Western Australia

This article is part of ourreviews of AI research papers, a series of posts that explore the latest findings in artificial intelligence.

Labor- and data-efficiency remain two of the key challenges of artificial intelligence. In recent decades, researchers have proven that big data and machine learning algorithms reduce the need for providing AI systems with prior rules and knowledge. But machine learningand more recently deep learninghave presented their own challenges, which require manual labor albeit of different nature.

Creating AI systems that can genuinely learn on their own with minimal human guidance remain a holy grail and a great challenge. According to Sergey Levine, assistant professor at the University of California, Berkeley, a promising direction of research for the AI community is self-supervised offline reinforcement learning.

This is a variation of the RL paradigm that is very close to how humans and animals learn to reuse previously acquired data and skills, and it can be a great boon for applying AI to real-world settings. In a paper titled Understanding the World Through Action and a talk at the NeurIPS 2021 conference, Levine explained how self-supervised learning objectives and offline RL can help create generalized AI systems that can be applied to various tasks.

One common argument in favor of machine learning algorithms is their ability to scale with the availability of data and compute resources. Decades of work on developing symbolic AI systems have produced limited results. These systems require human experts and engineers to manually provide the rules and knowledge that define the behavior of the AI system.

The problem is that in some applications, the rules can be virtually limitless, while in others, they cant be explicitly defined.

In contrast, machine learning models can derive their behavior from data, without the need for explicit rules and prior knowledge. Another advantage of machine learning is that it can glean its own solutions from its training data, which are often more accurate than knowledge engineered by humans.

But machine learning faces its own challenges. Most ML applications are based on supervised learning and require training data to be manually labeled by human annotators. Data annotation poses severe limits to the scaling of ML models.

More recently, researchers have been exploring unsupervised and self-supervised learning, ML paradigms that obviate the need for manual labels. These approaches have helped overcome the limits of machine learning in some applications such as language modeling and medical imaging. But theyre still faced with challenges that prevent their use in more general settings.

Current methods for learning without human labels still require considerable human insight (which is often domain-specific!) to engineer self-supervised learning objectives that allow large models to acquire meaningful knowledge from unlabeled datasets, Levine writes.

Levine writes that the next objective should be to create AI systems that dont require manual labeling or the manual design of self-supervised objectives. These models should be able to distill a deep and meaningful understanding of the world and can perform downstream tasks with robustness generalization, and even a degree of common sense.

Reinforcement learning is inspired by intelligent behavior in animals and humans. Reinforcement learning pioneer Richard Sutton describes RL as the first computational theory of intelligence. An RL agent develops its behavior by interacting with its environment, weighing the punishments and rewards of its actions, and developing policies that maximize rewards.

RL, and more recently deep RL, have proven to be particularly efficient at solving complicated problems such as playing games and training robots. And theres reason to believe reinforcement learning can overcome the limits of current ML systems.

But before it does, RL must overcome its own set of challenges that limit its use in real-world settings.

We could think of modern RL research as consisting of three threads: (1) getting good results in simulated benchmarks (e.g., video games); (2) using simulation+ transfer; (3) running RL in the real world, Levine told TechTalks. I believe that ultimately (3) is the most importantthing, because thats the most promising approach to solve problems that we cant solve today.

Games are simple environments. Board games such as chess and go are closed worlds with deterministic environments. Even games such as StarCraft and Dota, which are played in real-time and have near unlimited states, are much simpler than the real world. Their rules dont change. This is partly why game-playing AI systems have found very few applications in the real world.

On the other hand, physics simulators have seen tremendous advances in recent years. One of the popular methods in fields such as robotics and self-driving cars has been to train reinforcement learning models in simulated environments and then finetune the models with real-world experience. But as Levine explained, this approach is limited too because the domains where we most need learningthe ones where humans far outperform machinesare also the ones that are hardest to simulate.

This approach is only effective at addressing tasks that can be simulated, which is bottlenecked by our ability to create lifelike simulated analogues of the real world and to anticipate all the possible situations that an agent might encounter in reality, Levine said.

One of the biggest challenges we encounter when we try to do real-world RL is generalization, Levine said.

For example, in 2016, Levine was part of a team that constructed an arm farm at Google with 14 robots all learning concurrently from their shared experience. They collected more than half a million grasp attempts, and it was possible to learn effective grasping policies in this way.

But we cant repeat this process for every single task we want robots to learn with RL, he says. Therefore, we need more general-purpose approaches, where a single ever-growing dataset is used as the basis for a general understanding of the world on which more specific skills can be built.

In his paper, Levine points to two key obstacles in reinforcement learning. First, RL systems require manually defined reward functions or goals before they can learn the behaviors that help accomplish those goals. And second, reinforcement learning requires online experience and is not data-driven, which makes it hard to train them on large datasets. Most recent accomplishments in RL have relied on engineers at very wealthy tech companies using massive compute resources to generate immense experiences instead of reusing available data.

Therefore, RL systems need solutions that can learn from past experience and repurpose their learnings in more generalized ways. Moreover, they should be able to handle the continuity of the real world. Unlike simulated environments, you cant reset the real world and start everything from scratch. You need learning systems that can quickly adapt to the constant and unpredictable changes to their environment.

In his NeurIPS talk, Levine compares real-world RL to the story of Robinson Crusoe, the story of a man who is stranded on an island and learns to deal with unknown situations through inventiveness and creativity, using his knowledge of the world and continued exploration in his new habitat.

RL systems in the real world have to deal with a lifelong learning problem, evaluate objectives and performance based entirely on realistic sensing without access to privileged information, and must deal with real-world constraints, including safety, Levine said. These are all things that are typically abstracted away in widely used RL benchmark tasks and video game environments.

However, RL does work in more practical real-world settings, Levine says. For example, in 2018, he and his colleagues an RL-based robotic grasping system attain state-of-the-art results with raw sensory perception. In contrast to static learning behaviors that choose a grasp point and then execute the desired grasp, in their method, the robot continuously updated its grasp strategy based on the most recent observations to optimize long-horizon grasp success.

To my knowledge this is still the best existing system for grasping from monocular RGB images, Levine said. But this sort of thing requires algorithms that are somewhat different from those that perform best in simulated video game settings: it requires algorithms that are adept at utilizing and reusing previously collected data, algorithms that can train large models that generalize, and algorithms that can support large-scale real-world data collection.

Levines reinforcement learning solution includes two key components: unsupervised/self-supervised learning and offline learning.

In his paper, Levine describes self-supervised reinforcement learning as a system that can learn behaviors that control the world in meaningful ways and provides some mechanism to learn to control [the world] in as many ways as possible.

Basically, this means that instead of being optimized for a single goal, the RL agent should be able to achieve many different goals by computing counterfactuals, learning causal models, and obtaining a deep understanding of how actions affect its environment in the long term.

However, creating self-supervised RL models that can solve various goals would still require a massive amount of experience. To address this challenge, Levine proposes offline reinforcement learning, which makes it possible for models to continue learning from previously collected data without the need for continued online experience.

Offline RL can make it possible to apply self-supervised or unsupervised RL methods even in settings where online collection is infeasible, and such methods can serve as one of the most powerful tools for incorporating large and diverse datasets into self-supervised RL, he writes.

The combination of self-supervised and offline RL can help create agents that can create building blocks for learning new tasks and continue learning with little need for new data.

This is very similar to how we learn in the real world. For example, when you want to learn basketball, you use basic skills you learned in the past such as walking, running, jumping, handling objects, etc. You use these capabilities to develop new skills such as dribbling, crossovers, jump shots, free throws, layups, straight and bounce passes, eurosteps, dunks (if youre tall enough), etc. These skills build on each other and help you reach the bigger goal, which is to outscore your opponent. At the same time, you can learn from offline data by reflecting on your past experience and thinking about counterfactuals (e.g., what would have happened if you passed to an open teammate instead of taking a contested shot). You can also learn by processing other data such as videos of yourself and your opponents. In fact, on-court experience is just part of your continuous learning.

Ina paper, Yevgen Chetobar, one of Levines colleagues, shows how self-supervised offline RL can learn policies for fairly general robotic manipulation skills, directly reusing data that they had collected for another project.

This system was able to reach a variety of user-specified goals, and also act as a general-purpose pretraining procedure (a kind of BERT for robotics) for other kinds of tasks specified with conventional reward functions, Levine said.

One of the great benefits of offline and self-supervised RL is learning from real-world data instead of simulated environments.

Basically, it comes down to this question: is it easier to create a brain, or is it easier to create the universe? I think its easier to create a brain, because it is part of the universe, he said.

This is, in fact, one of the great challenges engineers face when creating simulated environments. For example, Levine says, effective simulation for autonomous driving requires simulating other drivers, which requires having an autonomous driving system, which requires simulating other drivers, which requires having an autonomous driving system, etc.

Ultimately, learning from real data will be more effective because it will simply be much easier and more scalable, just as weve seen in supervised learning domains in computer vision and NLP, where no one worries about using simulation, he said. My perspective is that we should figure out how to do RL in a scalable and general-purpose way using real data, and this will spare us from having to expend inordinate amounts of effort building simulators.

See the article here:
Reinforcement learning for the real world - TechTalks