OpenAI is adding support for plug-ins to ChatGPT — an upgrade that massively expands the chatbot’s capabilities and gives it access for the first time to live data from the web.
Quote:
This experimental feature is obviously similar to Microsoft’s Bing, which has custom tech that feeds GPT-4 (the language model underlying ChatGPT) information from the internet. However, OpenAI’s plug-in doesn’t just retrieve real-time information. It can also tie into APIs, letting it “perform actions on behalf of the user,” according to the company’s documentation. That could make it much more powerful — Bing could help you plan a vacation by telling you about flights and hotels, but ChatGPT could help you book it.
There are some obvious safety and security concerns with letting ChatGPT take actions on behalf of a user rather than just giving them information. Experts have already expressed worry about this in reaction to an experiment OpenAI conducted with GPT-4. When directed by a human tester, for example, the bot was able to generate instructions to employ a worker from TaskRabbit to complete a CAPTCHA it was unable to answer.
I'm not so sure this is a great idea. Beyond the obvious "I gave ChatGPT my credit card and it booked me a 5 star hotel in London, please refund me" issues, you may also give hackers a dream tool. We know it can write code, so what is to stop it from being tricked into hacking websites, or governments, or mass data gathering, or spamming forum sign ups. It's not the tool that is the issue, it's that people are #######s, and it's a pretty powerful tool. I guess this was inevitable.
I'm not so sure this is a great idea. Beyond the obvious "I gave ChatGPT my credit card and it booked me a 5 star hotel in London, please refund me" issues, you may also give hackers a dream tool. We know it can write code, so what is to stop it from being tricked into hacking websites, or governments, or mass data gathering, or spamming forum sign ups. It's not the tool that is the issue, it's that people are #######s, and it's a pretty powerful tool. I guess this was inevitable.
I have family that works in network security for Microsoft, so to say they and everyone they hang around with is a high-level performer would be an understatement. When ChatGPT launched several of her colleagues had full-on breakdowns, so I can only imagine what this news is doing to them. Things are happening so quickly that every day there's an overwhelming amount of info to process.
The Following User Says Thank You to Russic For This Useful Post:
I have family that works in network security for Microsoft, so to say they and everyone they hang around with is a high-level performer would be an understatement. When ChatGPT launched several of her colleagues had full-on breakdowns, so I can only imagine what this news is doing to them. Things are happening so quickly that every day there's an overwhelming amount of info to process.
I work in ML/AI at a big tech company and I have the opposite experience. The people that I work with are more excited than worried. My take on the plugins for GPT is mostly optimistic - it'll be able to deliver more up-to-date information about certain topics. Like google assistant but better.
The Following User Says Thank You to wireframe For This Useful Post:
I work in ML/AI at a big tech company and I have the opposite experience. The people that I work with are more excited than worried. My take on the plugins for GPT is mostly optimistic - it'll be able to deliver more up-to-date information about certain topics. Like google assistant but better.
Likewise, I'm thrilled about GPT and Copilot, the possibilities are nearly endless for the product portfolio I manage.
If you have ideas and sometimes struggle to put pen to paper, or express your ideas in compelling ways, GPT/Copilot is going to make you look like a rockstar. If you have no ideas, well, that's a problem. ChatGPT can only take you so far, it isn't independently thinking of new stuff that isn't already out there.
__________________
-James
GO FLAMES GO.
The Following User Says Thank You to TorqueDog For This Useful Post:
So, we're all basically going to be the fat people in Wall-E soon, right?
I was applying for jobs in a new career direction I'm taking and I asked GPT4 to write my cover letter. It blew my mind. I made a few changes to it, but damn. It was good.
So, we're all basically going to be the fat people in Wall-E soon, right?
I was applying for jobs in a new career direction I'm taking and I asked GPT4 to write my cover letter. It blew my mind. I made a few changes to it, but damn. It was good.
Ha. I've been saying that to my kids since that movie came out.
But only if we can keep dominance over the ai. If not it's Skynet time. Who's going to be the real John Connor?
So, we're all basically going to be the fat people in Wall-E soon, right?
I was applying for jobs in a new career direction I'm taking and I asked GPT4 to write my cover letter. It blew my mind. I made a few changes to it, but damn. It was good.
It's just going to make things like cover letters obsolete. They already are to a degree. The last time I was in the job market, a lot of online submissions specifically said not to submit a cover letter. They just wanted resumes with facts, then you come in to write a test or get an assignment to complete and submit. I think a lot of employers were already coming to the conclusion that cover letters were BS for the most part.
__________________
"A pessimist thinks things can't get any worse. An optimist knows they can."
I work in ML/AI at a big tech company and I have the opposite experience. The people that I work with are more excited than worried. My take on the plugins for GPT is mostly optimistic - it'll be able to deliver more up-to-date information about certain topics. Like google assistant but better.
While I agree that there's probably more reason for optimism than pessimism, if you work in AI at a big tech company, it would be weird to see a lot of fear around it. When your concern is keeping hackers from bringing down power grids, I can understand how the general mood can shift. Security people are in a constant arms race and this probably complicates things that are already very complicated.
That said, a few of the people who were panicking are now fully on board now that they've made tools that could make them a lot of money. Funny how that works.
IMO, AI is the next power tool of the construction age. Perhaps the digital power tool of the construction age. It allows people to do things inhumanly faster than they'd be able to do previously with manual tools.
That being said, I feel like in the longer term AI is a symbiotic relationship rather than a relationship to fear... but eh, others have beaten that sci fi theory to death. It's akin to the original worry about self checkout and automation making people lose jobs. Yes... it did. And then we figured out ways to address that and then we realized that automation was missing certain features of the old school method as well.
While I agree that there's probably more reason for optimism than pessimism, if you work in AI at a big tech company, it would be weird to see a lot of fear around it. When your concern is keeping hackers from bringing down power grids, I can understand how the general mood can shift. Security people are in a constant arms race and this probably complicates things that are already very complicated.
That said, a few of the people who were panicking are now fully on board now that they've made tools that could make them a lot of money. Funny how that works.
You're right that my work suggests that I'd be optimistic. ChatGPT still doesn't worry me very much on the security side. I guess you could connect a chatGPT plugin to Metasploit and have a slightly more convenient way to find known vulnerabilities. The biggest change will be in social engineering, which is probably the most important attack vector right now anyway. Instead of manually crafting a phishing email, you could generate one automatically. You'd still have to builds the exploits and find the targets but you could save some time writing the emails.
The bad news is that most computer systems are insecure because their owners don't secure them properly. Tools already exist to find and exploit these automatically.
The Following User Says Thank You to wireframe For This Useful Post:
The education world is going nuts over ChatGPT and other AIs. We're both really excited and really worried.
We're excited because this tool is obviously going to become something that makes a lot of tasks easier. I'm already using it to generate lesson ideas, streamline things, make writing more simple for students to read & understand, etc.
On the other hand, so much of education asks students to demonstrate their learning through written work. As a teacher, I now have to think carefully about tasks I can ask students to perform where they can show me what they have learned without the support of an AI. This is somewhat the same as the old issue of "how do I know if this is the student's work, or their work supported by a tutor / parent" except now it's so, so much easier and faster for kids to go to the AI.
An interesting consequence is that students - in the short term - are going to be doing a LOT more handwritten assignments. As educators and institutions get up to speed with AI, what it can and can't do, and understanding how it can be used to support thinking instead of replace it, a simple solution is having students write things in class, on paper.
Of course, handwriting instruction has more or less completely vanished over the last 20-30 years, so now I have to fight my way through atrocious handwriting to understand what my students are trying to say, and when I have meetings with elementary teachers I'm asking them to dedicate time to ancient handwriting drills.
Not something I expected to be asking for in 2023.
The Following 4 Users Say Thank You to driveway For This Useful Post:
It will be so fascinating to see where this goes for education. On one hand, graduates in 2035 will be the first ones who've had all 12 years with a robot tutor available 24/7. Will they have cheated, leaving them laughably unready for the world? Or will they have leveraged it to be far more effective?
One laughable difference between teacher's education and the real world was that in the education stage there's a ton of attention paid to catering learning to each individual's learning style.
This is all well and good until you enter an actual classroom and the realities of time get in the way. AI might actually be able to solve this problem, where concepts can be explained to each individual child slightly differently depending on their style and interests. "Explain the concept of natural selection, but pretend you're explaining it to somebody who only understands things in terms of the WWE/Pokemon/The Marvel Cinematic Universe."
The Following 2 Users Say Thank You to Russic For This Useful Post:
One laughable difference between teacher's education and the real world was that in the education stage there's a ton of attention paid to catering learning to each individual's learning style.
"Learning styles" are a myth and modern teacher education reflects this understanding. Unfortunately it is a powerful myth that did hold sway in teacher education for several decades. Currently, it is well known that there is no science to support the notion that people have inherent 'learning styles' through which they learn best, however it does remain an extremely popular idea.
Citations below the spoiler, if you're interested.
Spoiler!
Quote:
Study after study has shown that matching instructional mode to a student’s supposedly identified “learning style” does not produce better learning outcomes. In fact, a student’s “learning style” may not even predict the way they prefer to be taught or the way they actually choose to study on their own (Newton & Salvi, 2020).
Simply put, students’ learning preferences as identified via questionnaires do not predict the singular, best way to teach them. A single student may learn best with one approach in one subject and a different one in another. The best approach for them may even vary day-to-day. Most likely, students are best served when a variety of strategies are employed in a lesson. https://educationonline.ku.edu/commu...s-need-to-know
Quote:
Although the literature on learning styles is enormous, very few studies have even used an experimental methodology capable of testing the validity of learning styles applied to education. Moreover, of those that did use an appropriate method, several found results that flatly contradict the popular meshing hypothesis. We conclude therefore, that at present, there is no adequate evidence base to justify incorporating learning-styles assessments into general educational practice. Thus, limited education resources would better be devoted to adopting other educational practices that have a strong evidence base, of which there are an increasing number. However, given the lack of methodologically sound studies of learning styles, it would be an error to conclude that all possible versions of learning styles have been tested and found wanting; many have simply not been tested at all. Further research on the use of learning-styles assessment in instruction may in some cases be warranted, but such research needs to be performed appropriately. https://pubmed.ncbi.nlm.nih.gov/26162104/
Quote:
A recent review of the scientific literature on learning styles found scant evidence to clearly support the idea that outcomes are best when instructional techniques align with individuals’ learning styles. In fact, there are several studies that contradict this belief. It is clear that people have a strong sense of their own learning preferences (e.g., visual, kinesthetic, intuitive), but it is less clear that these preferences matter. https://www.scientificamerican.com/a...arning-styles/
Quote:
research surrounding learning styles is sparse and sometimes critical. It’s true that many students have a preference for one of the four learning styles. As of yet, however, no link has been found between a preferred learning style and academic achievement.[5]
This means that while students may have personal preferences, they don’t actually learn better using one style over another.[1] Because the theories have been around for decades and still fail to produce a measurable effect, an increasing number of educators question the extent to which they should deliberately use them in class.[7]
Rather than an educational theory, some researchers view learning styles as more of a personality test.[2] It can be useful to know which learning medium your students like the most to help them engage more with your lessons. But if you’re looking for a theory that models how the brain actually works, learning styles may not be it. https://www.waterford.org/education/...styles-theory/
Quote:
Many people, including educators, believe learning styles are set at birth and predict both academic and career success even though there is no scientific evidence to support this common myth, according to new research published by the American Psychological Association.
Previous research has shown that the learning styles model can undermine education in many ways. Educators spend time and money tailoring lessons to certain learning styles for different students even though all students would benefit from learning through various methods. Students study in ways that match their perceived learning style even though it won’t help them succeed. Some teacher certification programs incorporate learning styles into their courses, which perpetuates the myth for the next generation of teachers. Academic support centers and a plethora of products also are focused on learning styles, despite the lack of scientific evidence supporting them. https://www.apa.org/news/press/relea...ng-styles-myth
Quote:
Endless potential frameworks for categorizing learning styles exist, but the most popular one divides students into three types: visual, auditory, and kinesthetic learners. According to this theory, a self-reported visual learner learns best through visual content, while an auditory learner finds visual content less helpful than auditory material.
Yet the overwhelming consensus among scholars is that no scientific evidence backs this “matching” hypothesis of learning styles (Kirschner 2017, Pashler 2008, Simmonds 2014). While all learners can develop subjective preferences for studying or digesting material, studies deny that students learn better through a self-reported learning style. Instead, scholars increasingly call for educators to replace ‘neuromyths’ with resources and strategies rooted in evidence from cognitive and adult learning theory.
The Following 2 Users Say Thank You to driveway For This Useful Post:
I literally had GPT 4 completely rewrite one of my website site pages for SEO purpose, and had it write a brand new one from scratch with a handful of info, it game me the JSON schema for both, and it it was done a near full redesign in 10 minutes. The most time spent was font styling. I will check back in a month on how much impact it did to my ranking.
I also did the 100$ budget and get rich quick hack and it had build the code of the starting base of for a fiver like app using Javascript (legitimate code too)
Last edited by Firebot; 03-30-2023 at 09:36 AM.
The Following 2 Users Say Thank You to Firebot For This Useful Post:
"Learning styles" are a myth and modern teacher education reflects this understanding. Unfortunately it is a powerful myth that did hold sway in teacher education for several decades. Currently, it is well known that there is no science to support the notion that people have inherent 'learning styles' through which they learn best, however it does remain an extremely popular idea.
Citations below the spoiler, if you're interested.
Spoiler!
Ya you're right, I used the wrong terminolgy. What I meant was adjusting the content to subjects that the individual is passionate about (which is actually quite a bit more difficult that breaking it up into 4-5 styles).
Kinda like how when my kids were learning math, there were work books broken up by theme. Learn Grade 1 Math with Star Wars. Learn Grade 1 Math with Pokemon. Learn Grade 1 Math with Super Heroes. That sort of thing, only it could be done on the fly and each kid could can have their own interests woven into the content.
Kinda like how when my kids were learning math, there were work books broken up by theme. Learn Grade 1 Math with Star Wars. Learn Grade 1 Math with Pokemon. Learn Grade 1 Math with Super Heroes. That sort of thing, only it could be done on the fly and each kid could can have their own interests woven into the content.
I tried getting basic ChatGPT to write me an explanation of Natural Selection using Attitude Era WWE as an analogy, but it kept giving me an error message at the final step.
It is aware of the attitude era, and claims it can craft analogies for scientific concepts using popular culture, but trying to get it to craft this specific analogy appears to be beyond the free version's capabilities at this time.