221: Generative AI - Don’t Panic?
Вставка
- Опубліковано 28 лип 2024
- On today’s episode of Still To Be Determined we’re talking about Generative AI. It’s a little known topic, so maybe we’re the first place you’re hearing about it. Kidding aside … should we be freaking out about OpenAI?
Watch the Undecided with Matt Ferrell episode, AI Just Changed Everything … Again • AI Just Changed Everyt...
Join and support our UA-cam channel:
/ @stilltbd
Audio Podcast Version: www.stilltbd.fm/
Get in touch: undecidedmf.com/podcast-feedback
Support the show: pod.fan/still-to-be-determined
Follow us on X: @stilltbdfm @byseanferrell @mattferrell or @undecidedmf
Undecided with Matt Ferrell: / undecidedmf
00:00 - Intro & Feedback
09:11 - Generative AI Discussion - Наука та технологія
Yes! To solid state battery products deep dive.
Interested in advancements and positive news, plenty of places to hear about the negatives
For me the scariest thing about the race for AGI is that people are rushing to what they consider to be the finish line. And that Rush is not being governed by anyone other than the ones who are probably more motivated by money than anything else
Once I understood that the true power and purpose of business is to overcome barriers, I have found it difficult to agree with a company that wants to make it easier for them to do business. Creating an ethically sourced AI may be too expensive, and not worth it for one group, but absolutely worth it for another. There is no good reason to allow lives of the public to be negatively effected to make it easier for a company to operate. That's the public overcoming a barrier for the company, rather than the company over coming barriers for the public.
I think society will look much different in 10 years specifically because of ai. Creative industries receive much of the attention, but the wider effects will be profound. How ai will change the office of tomorrow is probably not the sexiest video topic, but this will be part of a major shift of how society functions. IT, software development, government and the insurance and medical industries, design and logistics. On and on. There is a lot of hype, because marketing found a new buzz word, but also because we have arrived at an inflection point.
I would love to see a series of a deep dive into the different aspects of AI. Good, bad and tbd (ha! see what I did?!) And yes, as a resource, I trust your channel(s) to handle something that is technological, even its policies, or lack thereof. Turn your headlight on and hit that rabbit hole! Thanks for all that you do.
and I wrote that before watching the whole video. One of the problems in our elected officials knowing what do... they barely know how to handle internet rules/regs, and the general public have had internet for 30ish years. By the time they get to AI, the internet will have proved to be the fad that the nay-sayers said it was going to be. Good luck in that rabbit hole, don't hit your head!
and until the end... just to reiterate, the good, the bad, and the tbd. (I couldn't help myself!)
I prefer to hear encouraging news. There is enough negativity all around. Please let us know where ai has solved problems in the real world.
The example with X-Ray: even AI bases it's predictions on human input, but medical science is not an exact science...
Forgery is also not new just more advanced than it was most criminals are on the cutting edge of how to break the idea in place.
I think the first thing we should be educating the public on is "what version or iteration of AI are we involved with?". Being there are 2 specific types, narrow and general, and each has a different function.
Narrow is what we are dealing with now. General is what people THINK we are building, the Skynets, the Asimov iRobots etc and I think first and foremost that distinction is very important. If that information was provided I believe, though I'm sure naively, people might step back and breathe. We could stop misinforming the public that we are "doomed" and our overlords are on the way. That does not mean we then can just sit back and do nothing because there are still the problems we are dealing with regarding the current state of AI and its use.
I don't know if Matt or anyone else watches Last Week Tonight but, John Oliver did a wonderful piece that I think ties in nicely with the Undecided video.
If I had to describe to someone who is not tech savvy what AI is at this moment I would say, "AI is currently like a newborn." As parents we supply "data" to our offspring so that they can function and learn while they progress through their life stages. LLMs are exactly what we are to those living beings; the parent(s). Data is scraped from the web, illegally and that is not even in dispute, and fed or taught to the AI. The AI then takes the overall avg of that data and attempts to predict X response. IE Ever have a young toddler standing around and they hear and adult curse? What does that child do? They end up repeating what they heard without any understanding of what it is they A ) heard nor B ) the implications of regurgitating that information. That is what AI is. Regurgitating information without truly understanding what that info is. It can only "understand" if it is given parameters to follow just like our cursing toddler. That child is told, "nooo. we don't say that!" and after confusion and possibly a punishment that child "learns" this is a bad thing to do. AI DOESN'T DO THAT! It doesn't do that because human beings, you know, those fallible creatures who created AI, are the ones in charge of teaching. Just like computers are still only as good at doing X job if the operator, a human being, is inputting the correct information in order to achieve X result. A computer never does anything on its own. It requires input of some kind in order to function properly.
To be honest there is so much to discuss and I already have provided a wall of text. The last thing I will say is politicians for the most part are not "tech savvy". So their input would be no different than having an upper level executive who doesn't know tech running a tech department and making poor choices for tech because they literally have zero understanding of it. They make poor choices based off of old or irrelevant information OR they are too finance/business minded to make proper decisions. How are we supposed to expect non tech enabled politicians to decide on how to handle any issue if it is something they have literally ZERO knowledge of? They are hearing and reading the same things the majority of the world hears, misguided if not completely misinformed information and making choices and comments about it. And yes I'm sure they have advisors however, that does not by any means guarantee they still really understand or if they care to.
Google is a major donor to the EFF
Thanks!
Thanks so much!
Ai jobloss? Maybe. But with swell robotics everywhere, Ai jobloss is the only thing I worry about anymore. Anyone else feel the same? Should we cease Ai?
If we want people in charge of the country that are ready and able to handle this new technological future we have coming at us we need to change how we elect them in the first place .
For example, in Oregon this next election we are going to be voting on whether or not to implement ranked choice voting throughout the state.
I for one am voting for it. We already have it in my county and I'm looking forward to actually starting to vote for people instead of against people .
Next, as far as AI is concerned and handling the new economy of the future, I think that we need to start pushing for these companies to use a certain portion of their compute to do things like work out the energy problem. Finally figure out cheap renewable, clean energy in massive amounts so that we can handle both the new technology and lower cost for everybody as jobs start going away .
That's just a taste of what I think we should do but we should expand it from there
Guns and AI comparison might be a good point of view.
Guns can be used just for fun (trap & skeet; target ranges…) or to get a job done (wildlife and forest preservation, taking down I’ll animals…) or to do harm (kill or harm others, blackmail, overtaking political systems by force…).
Same applies for AI.
To understand how to regulate either of them you need to understand how they work , how they can be used to do good and how the power they hand to the one controlling them can do harm. Then, and only then, you may stand a chance to control them to do more good than harm.
Unfortunately there are countries that believe that unlimited access to the source of arms is the one and only way to protect society. There is a big chance that like for guns this way of thought may prevail because it is backed both from those individuals that truly believe in this thought as well as by those that are invested in this technology….
Using AI to detect AI, that didn't work so well for the Texas A&M Professor, perhaps it has improved in the last year, but my recollection is that it misidentified a ton of stuff and feeding back his own stuff showed he was "guilty" as well. I know 12 months is a lifetime here, but not that long ago AI didn't detect AI well!
That professor used ChatGPT to see if ChatGPT was used. That was the failure in that particular case.
There is actual AI software available that checks specifically for plagiarism and AI tools that could have written the entire assignment(s). I took this directly from the article, "Platforms like Grammarly and Chegg also offer plagiarism-checking tools." Grammarly is $12 for premium which I would presume contains the tool that would check for plagiarism.
To be honest, maybe that professor should have just done their job and read each essay assignment vs asking AI to do the work for them. The irony is palpable.
The advent of AI helping to create misinformation isn't the problem. The underlying problem is a combination of the bias you mentioned along with a MASSIVE lack of critical thinking.
Simple things like looking at the timeliness of events, and realizing that "the story" that's being shoveled into the mouths of a certain demographic doesn't match the demonstrated, factual real timeliness.
We live in a world of Kellyanne Conway and "alternative facts."
Expecting people to think critically about something, when their identity and underlying value system requires that they DON'T think critically about whatever topic is at hand.
AI doesn't create this problem, but it sure as hell exacerbates the pre-existing problem.
The analogy would work better if instead of guns you paired AI with knives. Knives are tools that everyone can use for good reasons, but can be also weaponized. Guns are guns, you use them to kill or cause harm. Yeah there are legitimate uses for them, but 99% of the times you are better without them
made it about 60% of the way through this video and I'm just not into this channel. 90% of what's been said is meta discussion, almost nothing about AI has been said. lots of talk of cognitive bias followed by a regurgitation of corporate media propaganda points later in the video, very ironic and kind of sad to watch. comparing AI to weapons because it can "produce misinformation", silly. should pencils, computers, phones, etc. be regulated because people can and do produce misinformation with them? the argument falls to pieces under basic scrutiny, but you're more interested in pushing a corporate talking point because rich people fear guns.