Good side of AI

[Bigjay73]

Seems we only hear the fears about A.I., not that they aren't warranted. What's some things that we think A.I. can improve upon? Imma start with AI controlled traffic lights. No more 20 cars sitting at a red light while not a single car is utilizing the green light. So much wasted time and gas due to poorly programmed traffic signals

 

[SilentBirdHunter]

Most modern vehicles have AI...TPS, PCM, safety features, etc.

AI has been around since at least the 1950s when the Logic Theorist was sparked by various moments of pure scientific epiphany between the three collaborators, notably Herbert Simon, Allen Newell and Cliff Shaw.

The recent resurgence of AI in the news has been due to the strike of the
Writers Guild of America and the Screen Actors Guild last year
with concern that AI would replace workers in that profession.

 

[Ben Lamb]

Seems we only hear the fears about A.I., not that they aren't warranted. What's some things that we think A.I. can improve upon? Imma start with AI controlled traffic lights. No more 20 cars sitting at a red light while not a single car is utilizing the green light. So much wasted time and gas due to poorly programmed traffic signals

traffic lights are generally timed for overall flow of traffic, not congestion or lack of use at one light.

My SIL who is in academia says they use it constantly and it's been a boon for learning. Not sure how, but she's a Dr so smarter than me. AI for communication & translation (aka the babblefish from Hitchhikers Guide) is close to being mainstream, self driving cars, etc. Some cool stuff you can do w/AI to make life better for people.

The human condition & our collective history tells us that new technology will be used for both good and bad, as has James Cameron.

 

[buckbull]

For AI to work it needs a good source of data to learn from. I think law and financial market work would be ripe for AI usage. Could maybe reduce the number of paralegals and stock brokers.

 

[Nameless Range]

What is or isn't AI depends on the definition you're starting from. Some things I have used it for:

I've used AI (LLMs) to write code in 15 minutes that would've formerly taken me hours.
I work with folks who have used organizational policy/content/documents as training material for an AI, and in turn that trained chatbot instantly does a better job answering most obscure questions about organizational policy than any person alive.
My Ring Cameras only notify me when the AI within them sees a person on my doorstep - not deer, cars, or trampolines blowing by.
I use AI on aerial imagery for object detection and classification - a huge timesaver relative to the old ways.

Some things I hope AI can do:

-Take into consideration your genetics, habits, environment, work, family dynamics, etc - and give to you the best possible plan to do things like quit an addiction, be healthier, be happier, be a better parent, etc.
-Be better at diagnosing and prescribing than any meatspace doctor alive
-Allowing some of those with disabilities to functionally not have them
-Creating new tastes, smells, sounds never before heard
-Deciphering the language of other species of animal

What I fear it is most likely to do, maybe even by incorporating the levers above:

-Be better at hijacking your dopamine receptors, and thus your attention, and thus controlling you, than anything yet to exist.

 

[ovis2]

As discussed above, AI has really been around for a long time. In my academic world, we've variously called it iterative machine learning or even just computer modeling. I also tend to think there will be a lot of very positive applications of the evolving use of AI. The downside seems to be the purposeful manipulation of public opinions which, I hope, we will be able to figure out how to disrupt those manipulations - maybe through the help of AI .

 

[Jamen]

I am part of a project where this year we are going to have drones fly over fields and take high resolution pictures of these fields. AI will be used to help identify specific weeds. Goal would be if they can properly ID these weeds have that info shared with drones sprayers that will then be sent out and spot spray those areas. This was tried several years back but didnt gain much traction because the amount of data to process was so large. Drones and AI have come a long way in just a few years. I am optimistic that we will hopefully have some success.

 

[BEvans]

As a college student, I use ChatGPT every day. It can create five different reports with the same data when given the right prompts. It also is capable of writing perfect programming code. One of the things I like most about it is that it can chat like a person. If I have a question that's too long or hard to search online, I can ask ChatGPT and keep talking until I get an answer. One useful thing everyone here can do with it is write custom emails to send to our representatives about wildlife management. I can copy an entire article into the prompt and tell it what stance I want to take, and I'll have an email ready to send in 10 seconds.
And this whole comment was written by ChatGPT.

 

[NearerTheBone]

I work for a small geospatial data mapping company and "AI," more specifically Deep Convolutional Neural Networks for image segmentation, has figured heavily in the work we do (classifying features, such as trees or land cover, from high-res aerial imagery). In fact, it's completely revolutionized the scale and accuracy that we can map these features compared to older image segmentation and ML methods. We can produce datasets that just a few years ago would have been pure fantasy. So in that respect, AI has been a huge boon for our industry and end-users of these datasets.

Granted this is a small niche in the overall application and deployment of AI technologies, but these technologies are revolutionizing industries in so many sectors. The one thing that gets overlooked in this whole AI debate is the incredible volumes of data that are required to train robust models, which frankly makes LLMs like ChatGPT so impressive. Speaking from my own experiences working with DCNN models, we invest tons of time and effort in creating the data used to train the AI. I think it's easy to forget that these are human-devised systems, after all.

 

[LuketheDog]

As a college student, I use ChatGPT every day. It can create five different reports with the same data when given the right prompts. It also is capable of writing perfect programming code. One of the things I like most about it is that it can chat like a person. If I have a question that's too long or hard to search online, I can ask ChatGPT and keep talking until I get an answer. One useful thing everyone here can do with it is write custom emails to send to our representatives about wildlife management. I can copy an entire article into the prompt and tell it what stance I want to take, and I'll have an email ready to send in 10 seconds.
And this whole comment was written by ChatGPT.

No, he asked for things that AI will improve upon, letting it do all the thinking is (mostly) not an improvement...

 

[Johnny Bravo]

I would love to see it get rid of lawyers....

 

[Ben Lamb]

I would love to see it get rid of lawyers....

It's been made clear that the only people this will displace are those who don't want to be grist for the mill, or the milners.

 

[fleabagmatt]

new technology will be used for both good and bad, as has James Cameron.

I totally agree, James Cameron has for sure been used for good and for bad.

 

[Ben Lamb]

I totally agree, James Cameron has for sure been used for good and for bad.

they can't all be Dark Angel.

 

[Forkyfinder]

Seems we only hear the fears about A.I., not that they aren't warranted. What's some things that we think A.I. can improve upon? Imma start with AI controlled traffic lights. No more 20 cars sitting at a red light while not a single car is utilizing the green light. So much wasted time and gas due to poorly programmed traffic signals

Automated call systems. Low skill repetive work. All kinds of improvements - imagine your shopping cart just recognizing everything you put in it and giving you a total at the register.

 

[Elky Welky]

The manliest man to ever man:

 

[Ben Lamb]

I feel like a lot of you need to go live through the 80's to know why this is going to end poorly.

 

[BEvans]

No, he asked for things that AI will improve upon, letting it do all the thinking is (mostly) not an improvement...

I think letting ai do more thinking is an improvement. People cannot compete with the "thinking" of ai. I'm it will have its problems, but its either jump on the bandwagon and learn how to work with it or get left in the dust. Personally I'm excited by the prospect of not needing to be a rocket scientist to build a rocket.

 

[mxracer317]

For AI to work it needs a good source of data to learn from. I think law and financial market work would be ripe for AI usage. Could maybe reduce the number of paralegals and stock brokers.

Life savings isn’t something that many people are willing to trust to a computer system to solely manage. At least I’m not.

 

[SilentBirdHunter]

As discussed above, AI has really been around for a long time. In my academic world, we've variously called it iterative machine learning or even just computer modeling. I also tend to think there will be a lot of very positive applications of the evolving use of AI. The downside seems to be the purposeful manipulation of public opinions which, I hope, we will be able to figure out how to disrupt those manipulations - maybe through the help of AI .

USU is where I did my Ph.D. in the 1980s and I took a Artificial Intelligence course then. I loved Logan!