Did you know Neural is taking the stage this fall ? Together with an amazing line-up of experts, we will explore the future of AI during TNW Conference 2021. Secure your online ticket now !

The race to get humans to Mars is heating up. China recently joined the USA in targeting t he 2030s for crewed missions to the planet, but both their plans face enormous hurdles.

One of the main concerns is the threat posed by particle radiation from the sun, stars, and galaxies. These particles could increase the risk of cancer or cause acute radiation sickness during missions.

New radiation modeling suggests astronauts can be protected from these dangers — as long as their journeys are over within roughly four years.

Per the research paper :

The research combined geophysical models of particle radiation with simulations of how it would affect a spacecraft and its passengers.

The calculations suggest that the thickness of a spacecraft shield could play a key role on the trips. While relatively thick material may help protect astronauts, making it too thick could increase their radiation doses.

The researchers also estimated that the optimal timing for a flight to Mars would be when solar activity is at its peak.

The team believes a trip to Mars and back during this time is conceivable.

Yuri Shprits, a UCLA research geophysicist and co-author of the study paper, said the voyage could be done in under two years:

Greetings Humanoids! Did you know we have a newsletter all about AI? You can subscribe to it right here .

Artificial intelligence and the McData-fueled future of capitalism

Ba da ba ba bah , McDonald’s is capturing and storing biometric data on its customers without their knowledge or consent.

Per a report from The Register, McDonald’s may be facing a class action lawsuit after an Illinois customer sued the mega-corporation for allegedly violating the state’s Biometric Information Privacy Act (BIPA):

Illinois has some of the stiffest biometric privacy laws in the US.

The lawsuit apparently stems from the company’s use of automated drive-thru order takers in the form of chatbots.

Drive-thru customers were subjected to experimental natural language processing (NLP) AI in the state, in at least 10 of the company’s locations . While it’s unclear exactly what AI systems McDonald’s was using during the trial, it stands to reason the company would need to collect and store user data in order to train its AI.

The big picture

It’s hard to spot precedence in the wild, but there’s no denying the world sits on the rocky precipice of embracing autonomy. This very well could be the legal catalyst that kicks off the big business V big government debate over how we’re going to go about transitioning to the next technology paradigm for capitalism.

From a purely business-oriented POV, McDonald’s might not be in as bad a position as it appears. What’s an eight-figure lawsuit to company worth nearly $200 billion?

McDonald’s has been dabbling in AI systems for years now , and there’s an argument to be made that it’s poised to lead the charge when it comes to autonomous systems.

The perfect storm

Autonomous robotics technology is nothing new. Today it powers automotive factories and the garment manufacturing industry.

And that makes it easy for us to imagine other industries, such as fast food, adopting a similar approach. We’ve certainly heard a lot about burger-flipping robots and the end of entry-level jobs for the past decade.

The majority of discourse on automation focuses on the one-for-one human costs of replacement. We often envision the debate being about whether the efficiency and corporate labor cost reductions are worth the potential mass displacement of human workers.

But what if we stop thinking about McDonald’s like a greasy spoon and start thinking of it like Facebook, Google, or Microsoft.

The mainstream my recognize those as a social network, search giant, and OS developer respectively, but the truth of the matter is each one is an AI-first company. And with each passing year, AI endeavors make up a greater portion of their profits and net worth.

If McDonald’s were to convert its global market position as a restaurateur into a horizontal entry into the technology sector… interesting things could happen.

McDonald’s, but as an AI company

Strip away the what and how of where McDonald’s exists as a global corporation and you can compare it to other big tech businesses. The most apt comparison might be Facebook.

McDonald’s serves approximately one percent of the global population on a daily basis. Facebook, by contrast, reaches approximately 25% of the population . The biggest difference between the two, arguably, is that consumers typically have to pay to use the former’s services while Facebook monetizes its customers .

Let’s imagine a new McDonald’s where the food no longer costs money. Like Facebook, all you’d have to do is sign up and create a profile. Then, you could either go to a McDonald’s location to pick up food or request a delivery.

Every few orders, however, you may be asked to do something simple such as filling out a series of questionnaires similar to those “I’m not a robot” CAPTCHA’s where you click on the traffic lights or bicycles.

You might be tasked with ordering via voice or handwriting, so the system can capture your biometric data.

Most of the time, however, you’d just get free food for signing up and agreeing to McDonald’s terms and conditions.

Behold: Hypercapitalism

If this sounds a bit like socialism or communism, just remember: there’s no such thing as a free lunch. Whatever data McDonald’s could gather would be worth a fortune. It’s already a globally recognized brand with more than 38,000 locations in 100 countries.

The reason why so many big tech companies have pivoted to AI is because it’s a trillionaire’s market. Anyone can gather data, but only a few organizations have the money and infrastructure to gather data from billions of people at a time – and even fewer can ensure they’ll keep coming back for more no matter what.

There’s nothing stopping McDonald’s from using its burgers and nuggets to achieve the same goals as Facebook does with Candy Crush and conservative conspiracy theories.

The picture starts to come into focus when you consider that Facebook was founded in 2004 and it’s worth $280 billion while the first McDonald’s opened in 1955 and its only worth $170 billion.

Could McDonald’s turn feeding the hungry into the next big global data-gathering endeavor? What would you do for a “free” cheeseburger?

No. Just no. GPT-3 did not create a video game

GPT-3 is arguably the world’s most advanced text-generator. It was trained using supercomputing clusters, a nearly internet-sized dataset, and 175 billion parameters. It’s among the most impressive generative AI systems ever created.

But it absolutely did not create a video game.

You may have read otherwise. But it’s what you didn’t read that matters.

Background

GPT-3, for those who aren’t in the know, is a big powerful AI system that generates text from prompts.

At the risk of oversimplifying, you give GPT-3 a short input and ask it to generate text. So, for example, you might say “What’s the best thing about Paris?” and GPT-3 might generate text saying “Paris is known for its majestic views and vibrant night life,” and it’ll keep spitting out new statements every time you ask it to generate.

GPT-3 is pretty good at generating text that makes sense. So, if you were to keep generating new phrases based on the “What’s the best thing about Paris?” input, it’s likely you’ll get a bunch of different outputs that mostly made grammatical sense.

However, GPT-3 isn’t actually checking its facts or Googling things. It doesn’t have a database of verified information that it accesses before generating and injecting its opinion into things. It just tries to imitate the text its been trained on.

Without filtering, GPT-3 is as likely to say something xenophobic about Parisians and/or the French people as it is something positive. And, most importantly, it’s just as likely to say something factually incorrect.

GPT-3 does not think. It does not understand anything . It doesn’t know what a dog is, it can’t understand the color blue, and it has no mental capacity for continuity or sense. It’s just algorithms and computer tricks.

If you can imagine 175 billion monkeys banging on 175 billion typewriters you can imagine GPT-3 at work. Except, in GPT-3‘s case, instead of letters, the keys all have sentences and phrases on them. And for every monkey there’s a human standing there changing out text templates to fit specific themes.

What’s this about a video game?

A gambling website called, aptly, “ Online Roulette ” started sending PR emails out a couple weeks ago claiming that GPT-3 had created a video game.

Here’s the thing: This wasn’t a pitch for a game or even an AI-related pitch. It was a pitch for a survey about how gamers responded to a PR pitch for a game that doesn’t exist that referenced text that was generated by GPT-3.

So here’s a few things to keep clear:

I call bullshit

This isn’t to say GPT-3 can’t be involved in the development of a video game. AI Dungeon is a game that uses text-generating AI to create novel text-based game experiences. As anyone who’s played it can attest, it’s often cogent in a surreal way. But it’s just as often weird and nonsensical.

However, this marketing pitch from Online Roulette has nothing to do with the creation of an actual game.

Let’s start with the survey and work our way back. Here’s the “methodology” section on the website the original pitch refers to:

Who were these “avid” gamers? Were they Mechanical Turk workers? Were they Online Roulette customers? Were they Twitter respondents? We don’t know.

What exact images and text were the respondents exposed to? Because if they were exposed to this website , the one the above images came from, they weren’t exposed to the game GPT-3 supposedly spit out.

The entire website describes a game GPT-3 allegedly generated, but nowhere is GPT-3 quoted or is it made explicit that any of the text on the site is directly attributable to GPT-3.

Exactly what did GPT-3 generate?

Why weren’t the survey respondents told they were evaluating a game allegedly generated by an AI?

The real problem

For the sake of argument, let’s say the images and text on the Online Roulette website were actually spit out by GPT-3 in the form we see them. They weren’t. But let’s just say they were.

It would be useless information.

It’s insulting that anyone would think game development and design is such a whimsical field that a machine could randomly spit out ideas that could challenge human talent.

Game developers spend lifetimes learning the industry and its fans. It takes years to gain a perspective on the $90 billion video game market. And even if you have an amazing idea, that doesn’t mean it’ll translate into a compelling game.

Nobody is sitting around waiting for a random video game pitch-generator to spark their development careers.

If coming up with a good idea was the hard part, there’d be more game developers than there are players. That’s like saying GPT-3 is a threat to Metallica because it can write random lyrics about things that are dark.

But it’s even more insulting that, according to this article , at least one person involved with generating the data actually expects us to believe the results from GPT-3 weren’t cherry-picked. But that’s a ludicrous claim.

The reality of the situation

We don’t know who was surveyed and we don’t know what the respondents actually saw. We also don’t know what parts of the game’s marketing pitch GPT-3 actually generated.

For all we know, the “researchers” generated results until they found something they liked and then started using prompts specific to that result to generate dozens or hundreds of options from which they then curated, arranged, and edited to go along with the images their human artists created.

Basically, Online Roulette is asking you to believe that a sketchy marketing pitch with zero details, about a survey referencing a game that doesn’t exist, highlights an example of working artificial intelligence.

The only thing impressive about “Candy Shop Slaughter” is that we’re talking about it.