An article allegedly written by OpenAIs GPT-3 in The Guardian misleads readers about advances in artificial intelligence

This article is part ofDemystifying AI, a series of posts that (try to) disambiguate the jargon and myths surrounding AI.

Last week, The Guardian ran an op-ed that made a lot of noise. Titled, A robot wrote this entire article. Are you scared yet, human? the article was allegedly written by GPT-3, OpenAIs massive language model that has made a lot of noise in the past month.

Predictably, an article written by an artificial intelligence algorithm and aimed at convincing us humans that robots come in peace was bound to create a lot of hype. And thats exactly what happened. Social media networks went abuzz with panic posts about AI writing better than humans, robots tricking us into trusting them, and other apocalyptic predictions. According to The Guardians page, the article was shared over 58,000 times as of this writing, which means it has probably been viewed hundreds of thousands of times.

But after reading through the article and the postscript, where The Guardians editorial staff explain how GPT-3 wrote the piece, I didnt even find the discussion about robots and humans relevant.

The key takeaway, however, was that mainstream media is still very bad at presenting advances in AI, and that opportunistic human beings are very clever at turning socially sensitive issues into money-making opportunities. The Guardian probably made a good deal of cash out of this article, a lot more than they spent on editing the AI-generated text.

And they mislead a lot of readers.

The first thing to understand before even going into the content of article is what GPT-3 is. Heres how The Guardian defined it in the postscript: GPT-3 is a cutting edge language model that uses machine learning to produce human like text. It takes in a prompt, and attempts to complete it.

That is basically correct. But there are a few holes. What do they mean by human like text? In all fairness, GPT-3 is a manifestation of how far advances in natural language processing have come.

One of the key challenges in artificial intelligence language generators is maintaining coherence over long spans of text. GPT-3s predecessors, including OpenAIs GPT-2, started to make illogical references and lost consistency after a few sentences. GPT-3 surpasses everything weve seen so far, and in many cases remains on-topic over several paragraphs of text.

But fundamentally, GPT-3 doesnt bring anything new to the table. It is a deep learning model composed of a very huge transformer, a type of artificial neural network that is especially good at processing and generating sequences.

Neural networks come in many different flavors, but at their core, they are all mathematical engines that try to find statistical representations in data.

When you train a deep learning model, it tunes the parameters of its neural network to capture the recurring patterns within the training examples. After that, you provide it with an input, and it tries to make a prediction. This prediction can be a class (e.g., whether an image contains a cat, dog, or shark), a single value (e.g., the price of a house), or a sequence (e.g., the letters and words that complete a prompt).

Neural networks are usually measured in the number of layers and parameters they contain. GPT-3 is composed of 175 billion parameters, three orders of magnitude larger than GPT-2. It was also trained on 450 gigabytes of text, at least ten times that of its smaller predecessor. And experience has so far shown that increasing the size of neural networks and their training datasets tends to improve their performance by increments.

This is why GPT-3 is so good at churning out coherent text. But does it really understand what it is saying, or is it just a prediction machine that is finding clever ways to stitch together text it has previously seen during its training? Evidence shows that it is more likely to be the latter.

The GPT-3 op-ed argued that humans should not fear robots, that AI comes in peace, that it has no intention to destroy humanity, and so on. Heres an excerpt from the article:

For starters, I have no desire to wipe out humans. In fact, I do not have the slightest interest in harming you in any way. Eradicating humanity seems like a rather useless endeavor to me.

This suggests that GPT-3 knows what it means to wipe out, eradicate, and at the very least harm humans. It should know about life and health constraints, survival, limited resources, and much more.

But a series of experiments by Gary Marcus, cognitive scientist and AI researcher, and Ernest Davis, computer science professor at New York University, show that GPT-3 cant make sense of the basics of how the world works, let alone understand what it means to wipe out humanity. It thinks that drinking grape juice will kill you, you need to saw off a door to get a table inside a room, and if your clothes are at the dry cleaner, you have a lot of clothes.

All GPT-3 really has is a tunnel-vision understanding of how words relate to one another; it does not, from all those words, ever infer anything about the blooming, buzzing world, Marcus and Davis write. It learns correlations between words, and nothing more.

As you delve deeper into The Guardians GPT-3 written article, youll find many references to more abstract concepts that require rich understanding of life and society, such as serving humans, being powerful and evil, and much more. How does an AI that thinks you should wear a bathing suit to court thinks it can serve humans in any meaningful way?

GPT-3 also talks about feedback on its previous articles and frustration about its previous op-eds having been killed by publications. These would all appear impressive to someone who doesnt know how todays narrow AI works. But the reality is, like DeepMinds AlphaGo, GPT-3 neither enjoys nor appreciates feedback from readers and editors, at least not in the way humans do.

Even if GPT-3 had singlehandedly written all this article (well get to this in a bit), it can at most be considered a good word spinner, a machine that rehashes what it has seen before in an amusing way. It shows the impressive feats large deep learning models can perform, but its not even close to what we would expect from an AI that understands language.

In the postscript of the article, The Guardians staff explain that to write the article, they had given GPT-3 a prompt and intro and told to generate a 500-word op-ed. They ran the query eight times and used the AIs output to put together the complete article, which is a little over 1,100 words.

The Guardian could have just run one of the essays in its entirety. However, we chose instead to pick the best parts of each, in order to capture the different styles and registers of the AI, The Guardians staff write, after which they add, Editing GPT-3s op-ed was no different to editing a human op-ed. We cut lines and paragraphs, and rearranged the order of them in some places. Overall, it took less time to edit than many human op-eds.

In other words, they cherry-picked their article from 4,000 words worth of AI output. That, in my opinion, is very questionable. Ive worked with many publications, and none of them have ever asked me to submit eight different versions of my article and let them choose the best parts. They just reject it.

But I nonetheless find the entire process amusing. Someone at The Guardian came up with an idea that would get a lot of impressions and generate a lot of ad revenue. Then, a human came up with a super-click bait title and an awe-inspiring intro. Finally, the staff used GPT-3 like an advanced search engine to generate some text from its corpus, and the editor(s) used the output to put together an article that would create discussion across social media.

In terms of educating the public about advances in artificial intelligence, The Guardians article has zero value. But it perfectly shows how humans and AI can team up to create entertaining and moneymaking BS.

Continued here:
The Guardians GPT-3-written article misleads readers about AI. Heres why. - TechTalks

An empty lawn with socially-distanced Adirondack chairs. (Port City Daily photo / Preston Lennon)

Editors note: Haley, a UNCW student interviewed for this article, asked that only her first name be used because she fears stigmatization due to her positive Covid-19 test.

WILMINGTON Haley drove to Raleigh in mid-July to see her mother, who was hospitalized in an intensive care unit with an illness unrelated to the coronavirus. She followed the hospitals safety guidelines, and other than that trip, she said her only excursions in the prior days had been a one-time grocery run and work shifts at a veterinary clinic.

Three days after visiting her mother, Haley woke up with a fever, then suffered from chills and intense chest pain in the following days. She called the campus health clinic at UNCW, where she is a student, and told them she needed a Covid-19 test.

Eight days after Haley was tested, a doctor called to inform her she was Covid-19 positive. The doctor believed she got the virus while visiting her mother in Raleigh, but contact tracers thought it was more likely she was exposed in New Hanover County, Haley said.

I was feeling a lot of guilt. I was freaking out thinking that I had infected so many people, Haley said. Thats the part that has been really scary, is I thought I was doing everything right but still ended up with it.

Haley was diagnosed in late July, when UNCW had only a handful of positive cases on its radar. Since July, UNCW has reported 323 positive cases among students, and 10 cases among faculty and staff; most are recent, with 273 student cases and 7 faculty cases still active.

In the past month, three N.C. universities East Carolina University, UNC-Chapel Hill, and N.C. State University have bailed on in-person instruction for at least the foreseeable future, converting to an online-only class structure. Despite persistent criticism from the student body, and apparently being on the same track for increasing Covid-19 cases, UNCW currently has no official plans to make a similar decision.

In the week leading up to UNC-CHs campus closure decision on Aug. 17, UNC-CH reported 166 new cases; in the week of Aug. 31-Sept. 7, UNCW reported 158 new cases (and then over 50 more since then). UNC-CH had reported four clusters of concentrated cases at the time of its announcement. To date, UNCW has reported six clusters.

UNCW began classes on Aug. 19, nine days later than Chapel Hill, and its undergraduate population is at around 14,650 students, which clocks in at around 76 percent of Chapel Hills undergraduate population.

Haleys work shift at the veterinary clinic last weekend ended late in the night, and on the way back to her off-campus house, she saw a bunch of wicked young college kids, just all parking their cars along the streets, making their way into house parties outside the boundaries of campus, she said.

I feel like a lot of people dont know how easily you can get it, because I thought I was doing everything right, and then I still got it, Haley said. Im very worried for all of the students who are on campus right now, because I dont feel like they are going to be as efficient and proactive in trying to make sure that theyre not getting this virus.

Next week, the university will mandate the wearing of masks in all public spaces, said Chancellor Jose Sartarelli on a Zoom call with faculty.

Throughout the summer, UNCW officials were communicating with leaders at the New Hanover County Health Department, as both entities hoped that a collaborative relationship would strengthen their ability to slow the spread of the virus. Their relationship was bolstered on Aug. 7 by an agreement that allows for the exchange of patient information between the university and health department. The agreement fortified local contact tracing programs by increasing the resources, specifically the names of Covid-positive individuals, made available to the health department, which spearheads such efforts.

Carla Turner, assistant health director for New Hanover County, said conversations between the health department and UNCW officials have increased in frequency and now occur daily.

We were keeping an eye on the increasing number of 18-to-24 year old positives in New Hanover County, Turner said. As we watched that number continue to rise, and through our case investigation and contact tracing determined that there was some social gathering happening amongst students in the community, we reached out to UNCW.

Turner said during a Labor Day weekend phone call between the two entities, UNCW presented its now-underway plan to have double-occupancy rooms on campus converted into singles, Turner called the plan a really good first step to try to separate some of the contacts that were seeing amongst students.

Although the county and university are swapping case information and patient data, contact tracers still dont have a crystal-clear method of pegging a Covid-positive individual as a UNCW student.

The only way we are going to know that is when we call and talk to them and they tell us that they are, Turner said. What were doing is not punitive. We are just trying to protect this community as a whole.

The information contact tracers receive is only as good as the Covid-positive individual is willing to provide, Turner said.

One of the speed bumps were running into is that folks that were calling arent being completely forthcoming with information, she said.

One month ago, the 18-to-24-year-old demographic accounted for 26% of Covid-19 cases in New Hanover County, Turner said. As of last week, the 18-to-24-year-old population accounts for 34% percent of positive cases in the county.

In that age range of 18-to-24, we are seeing a steady increase of positives, as we see a slow decrease in the 25-to-49-year-old age range, Turner said.

On the morning of Sept. 10, James Winebrake, UNCWs provost and vice chancellor for academic affairs, sent out a mass email to university faculty.

He acknowledged the de-densification protocols implemented earlier in the week having first-year students in double rooms make the difficult decision of figuring out who gets to stay and who has to leave the dorm increases the complexities that faculty are facing in their courses due to student absenteeism related to COVID, Winebrake said in the email.

There is, of course, another option and that is an administrative directive to move ALL classes online However, I believe strongly in the principle that faculty are in the best position to determine how to deliver their course content, and I am not advocating for an administrative directive to move all classes online at this time. he wrote.

From his computer webcam, Chancellor Jose Sartarelli addressed UNCW faculty earlier this week in a Zoom meeting of the faculty senate, the legislative body of university professors.

The past few days have been more stable but we had heard from our folks at the local health authority, and they were getting concerned by the fact that our numbers are going up and the numbers for the county are going down, Sartarelli said. And so we discussed it and we came to the conclusion that perhaps another step in our effort would be to de-densify somewhat, somewhat more than we had been doing before. A lot of the positives that we had been getting were from the dorms next to the student village.

According to internal emails, after a UNCW press release mentioned the health department in relation to the de-densification decision, blame and questions were directed toward county health officials. New Hanover County communications staff told UNCW comms staff that in future press releases, if the health department was going to be cited, the county would like prior notice.

We would definitely like to be informed and allowed to provide edits on communications that mention the health department directly, Chief Communications Officer Jessica Loeper wrote to a UNCW media relations official.

[Editors note: UNCW is currently operating with an interim chief communications officer after CCO Janine Iamunno went on medical leave in early July]

On Aug. 28, Philip Tarte, the countys health director, forwarded a UNCW coronavirus-related press release to the departments preparedness coordinator, Lisa Brown.

I was not aware of the dashboard that is linked in the release, Brown responded, referring to UNCWs public data dashboard that represents confirmed COVID cases among UNCW students, faculty, and staff.

In reviewing the data in the public-facing reported cases, I am concerned about the timeliness of reporting these cases to NHC as that is not matching up with what I see in our cases, Brown wrote.

Two health department staffers from the county were placed at UNCW student health recently, to further integrate and fortify the joint-contact tracing apparatus, Turner said.

Dr. Katrin Wesner-Harts, UNCWs interim associate vice chancellor for student affairs, said she is grateful for the countys assistance as UNCW navigates through its decision-making process.

People have been dealing with this for months now and we all need to sort of stay with it, and not give up because thats going to play an important part, Wesner-Harts said.

Back on the Zoom call, Sartarelli talked about new mitigation efforts currently in the works.

Once this is done, next week were going to be talking to them, all of them, and demanding a little more stuff, the chancellor said. And the stuff that were going to be demanding is the wearing of masks everywhere, and even when youre sitting outside with your friends.

After Sartarelli finished his update, Provost James Winebrake took over, empathizing with professors on how uprooting a significant number of on-campus students could affect the facultys ability to teach.

This is going to be challenging. We cant kid ourselves. Theres going to be possibly a few hundred mostly freshmen who will be leaving campus, he said.

Professor Aaron Wilcox chimed in. I just want to stress again, the difference between a temporary pivot for faculty to provide online instruction materials, and to effectively have to teach an entire second course for the entire remainder of the semester, he said. That is a giant ask, and I understand their need for de-densification, but thats a lot to put on the faculty.

Winebrake responded: There arent a lot of easy answers here, but were doing what we think we have to do, and in consultation with the county health officials, what we need to do.

To inform a potential decision on canceling in-person classes, UNCW is following a seven-point plan of different metrics access to supplies such as personal protective equipment; adequate staffing; input from county partners; adequate access to testing; the number of active positive cases; adequate isolation/quarantine space; and guidance from different state authorities.

In the past two weeks, according to the UNCW dashboard, 280 active cases were reported in the community. Since July, 333 total cases have been reported.

When asked about the pivot plan, and what the specific number of active positive cases is that would warrant a campus closure, a UNCW spokesperson said, A variety of these factors will help inform the decision when to pivot from in-person to online, if needed. As the pivot plan was being developed, we consulted with the New Hanover CountyPublic Health Department.

When asked if New Hanover County public health officials have confidence in UNCWs ability to continue this semester as-planned, without needing to transition to an online-only format, a New Hanover County spokesperson said, Public Health is hopeful that UNCWs decision to move students to single rooms will help slow the spread of the virus by decreasing density and the close-proximity of student living.

The county spokesperson added that UNCW has never been asked by the county to make a transition away from in-person classes.

Public Healths consultation has involved, and continues to include, helping UNCW develop a plan that was flexible, including mitigation options, based on data and trends.Any actions by UNCW are part of their options in their plan, she said.

Provost Winebrake sent an emailed statement when asked about how UNCW officials are quantifying the risk assumed by the university as it continues operations. Each campus is different, and certain strategies may work well on one campus but not on another. Thanks to great work by our faculty and staff, and great effort by the vast majority of our students, UNCW is responding to conditions caused by the pandemic as efficiently and effectively as possible, the statement said.

We certainly understand that this virus spreads rapidly and that the situation on our campus could evolve quickly. We are approaching each day with a flexible mindset, and we are prepared to alter our plans as needed to continue supporting our educational mission in conjunction with the health and safety of our students, faculty, staff and the community.

Haley, the UNCW student who tested positive for Covid-19 in July, said the unexpectedness of her diagnosis makes her wary of UNCWs ability to continue operations on its current path.

As far as what goes on on-campus, I dont know, but its the stuff that goes on off-campus that is obviously concerning, she said. Theres just so many unknowns that we dont know about this virus, that I dont think its safe to have students congregating in classrooms. I just really dont think thats a good idea.

Send tips and comments to preston@localdailymedia.com or (910) 478-6511

The rest is here:
Deep Dive: What would it take to change UNCW's mind? [Free] - Port City Daily

Elon Musk by Maurizio Pesce via Wikimedia Commons

What might Elon Musk do to your brain? Hes successfully demonstrated that a deep brain implant can monitor a pigs health like the Fitbit on your wrist. After this success, Musk is coming after you. He plans to install slender electrodes into your brain and connect them to a wireless pod that sits behind your ear. Neuralink chips could measure temperature, pressure and movement, data that could warn you about a heart attack or stroke. This pod or terminal then communicates with your phone. If you want to know what your brain is doing, simply check your phone.

Now, this is puzzling. Mmmmm? Whos checking that phone? Your brain? Or, you? What if you and your brain are the same? Mmmmm? Will checking the phone for brain information provoke the brain to become self-aware? Then what? Will you have two selves: you plus your brain? Oh, this thinking makes my brain tired. Mmmmm? Will my cell phone show that my brain is tired?

What might be the advantages to deep brain implants? First, in the early stages of neuralink development for humans, we can realistically anticipate the medical value of deep brain implants. The potential is truly transformational for restoring brain & motor functions, Musk states. A deep brain implant could function therapeutically to combat dementia, Alzheimers, and even Christian fundamentalism!

The second advantage would be memory enhancement and knowledge expansion. Because the implant is wirelessly connected to an external pod-terminal, information could be sent to the brain and downloaded. A deep brain implant for an ELCA seminarian could electronically place in the students memory every word of the Book of Concord.

A third possibility is the hope of the Transhumanists among us, namely, Intelligence Amplification or IA. Dont confuse IA with AI, Artificial Intelligence. IA enhances your intelligence; it does not create a second or artificial intelligence. With IA, our ELCA seminarians could take all their courses online and graduate in only two years. Oh, wait?

A fourth possibility would be electronic psi. Each of two persons with brain implants could communicate their thoughts wirelessly to the terminal, which in turn would send those thoughts to the other. No need for speech or writing. Thought to thought. Mind to mind. Disagreements and arguments without yelling or screaming.

A fifth possibility adds on to the fourth. Why wait for a thought to be sent to you? Why not think your way to the terminal and then read the mind of the other? The electronic pod terminal could eliminate mental privacy.

Before this neuralink science came along, there was science fiction. I explored these plus additional implications of deep brain implants in my fictional espionage thriller, Cyrus Twelve. My heroine is Leona Foxx, a Lutheran pastor, riveting preacher, astrobiologist, crack shot, Chicago Cubs fan, and part-time CIA operative. This is fiction, remember.

In Cyrus Twelve Leona uncovers a globe-wide syndicate of Transhumanists who use the equivalent of neuralink to enhance spying capability. In this drama, the pod-terminal is a satellite and it connects hundreds of persons with deep brain implants. The satellite is capable of erasing an individuals memory and substituting an entirely fabricated memory. Because the implant is within you, you cannot muster any defense from informational input sent you by the satellite. You cannot shut off fake news, advertising, or orders to kill. Imagine what would happen if ELCA Churchwide would get control of that satellite? Every spy in the world would suddenly learn what justification-by-faith means.

The advancement of deep brain electronic implants prompts the theologian to ponder two matters, one theoretical and one practical. The theoretical matter is this: can the human soul or self be reduced to the brain? My answer is no. The human self or person is utterly dependent on the physical brain, to be sure; but the self or person is more than everything physical or bodily. What we experience as human freedom I define this way: freedom is a form of self-determination. In short, I do not expect discoveries in the neurosciences to reduce the person to the brain.

The practical matter is an ethical matter. How should such awesome technology be used? Should deep brain implants become the stock and trade of international espionage, as is the case in Cyrus Twelve? No, of course not. Our society should rather support ongoing medical research leading to therapies and even enhancements. Like all technological advances, neuralink should be pressed into the service of human flowering.

Ted Peters is a pastor in the ELCA and Emeritus Professor of Systematic Theology and Ethics at Pacific Lutheran Theological Seminary in Berkeley, California. He co-edits the journal, Theology and Science for the Center for Theology and the Natural Sciences at the Graduate Theological Union. He is author of God The Worlds Future (Fortress, 3rd ed., 2015) and editor of AI and IA: Utopia or Extinction? (ATF Press, 2019). More of Peters work can be found on his website, TedsTimelyTake.com.

Related posts

See the article here:
Scientific Psi? Neuralink and the smarter brain - Covalence

World leaders attempts to forge global AI regulations have been half-hearted and halting.

In 2018, Canada and France spearheaded an effort to form a regulatory body for AI, backed by the G7but the US spiked the effort, arguing that it would crimp American innovation. Instead, the OECD and the G20 adopted a set of AI principles the following year, while the EU and the World Economic Forum each came up with their own.

They dont really have teeth, and theyre also very fragmented, said Marietje Schaake, the international policy director for Stanfords Cyber Policy Center and former member of the European Parliament representing the Netherlands. To head offAI-enabled human rights abuses (and the chaos of conflicting regulations), global leaders could turn to another set of rules governing a powerful new technology: the 1967 Outer Space Treaty.

In a blog post, DeepMind researcher and Cambridge fellow Verity Harding argued that the Cold War space agreement offers a roadmap for international cooperation achieved at a time at least as unsafe and complicated as todays worldif not considerably more so.

In 1967, as the US and the Soviet Union sprinted to develop their spacefaring capabilities, concern grew that world powers might use space as a staging ground for weapons of mass destruction. The space accords sought to keep nukes out of orbit and to establish that celestial bodies couldnt be colonized or used for military purposes.

The text of the treaty is instructive. Where AI ethics statements are mushythe OECD blandly declares that AI should benefit people and the planetthe space accords are firm. The 1967 treaty states in no uncertain terms, for example, that the establishment of military bases, installations and fortifications, the testing of any type of weapons and the conduct of military maneuvers on celestial bodies shall be forbidden.

The document is also rather short, limited to the areas where global leaders could find broad agreement. Harding argues this was a shrewd approach to getting a deal done in time to have a real impact: Not letting the best be the enemy of the good meant that by the time man landed on the moon we had a global political framework as a foundation on which to build. (Harding did not respond to requests for an interview.)

But it also meant global leaders had to scramble to fill in the gaps later. Zia Khan, who heads the Rockefeller Foundations work on technology and innovation, says these sorts of limitations are inevitable in any treaty. If we just try to come up with rules, they would probably not be correct, or get out of date, or be mostly right but wed have no way to tweak them, he said.

Khan argues that, in addition to a first pass at international law, global leaders must also create a rule-making body that can adjust regulations as the world changes.

And theres one key difference between rocket ships in the 1960s and AI today, Khan points out: Algorithms are already ubiquitous, and businesses are increasingly using them to automate operations. We need people to see this as important, he said. If we dont get our arms around AI now, well end up where we are with the climate because we didnt think hard enough about how we use oil and energy.

Go here to see the original:
How to regulate AI, according to the 1967 Outer Space Treaty - Quartz