Community IT Innovators Nonprofit Technology Topics

Artificial Intelligence (AI), and Machine Based Learning: Nonprofit Tech of the Future pt 3

December 16, 2022 Community IT Innovators Season 3 Episode 54
Community IT Innovators Nonprofit Technology Topics
Artificial Intelligence (AI), and Machine Based Learning: Nonprofit Tech of the Future pt 3
Show Notes Transcript

In pt 3 of this interview series on high tech and nonprofits, we wrap up our discussion about cutting edge technology with 3 nonprofit tech experts with decades of experience and a deep interest in tech innovations; Johan Hammerstrom, CEO of Community IT Innovators, Steve Longenecker, our Director of IT Consulting, and Nura Aboki, Senior IT Business Manager and Consultant. 

Johan, Nura, and Steve look into applications of Artificial Intelligence for nonprofits, from a chatbot who can answer questions while taking donations to on-demand images created by AI for your website or brochure, crafted to appeal to universal emotions. Nura also explores machine based learning and how forward looking nonprofits may be able to use this tech to save people from repetitive tasks in the future.

Is your nonprofit forward looking? Are you wondering what tech changes your nonprofit will be facing in the next 3-5 years and how embracing new technologies could enhance and support your nonprofit mission and goals?

Join us for a quick preview of where we think these technologies are going and how we expect this trendy tech to impact the nonprofit sector. 

In parts 1 and 2 of this series our experts discussed Virtual Reality and Crypto. You can find those episodes in your subscription feed or on our website.

_______________________________
Start a conversation :)

Thanks for listening.


Carolyn Woodard:  Welcome. My name is Carolyn Woodard, and I am the Director of Outreach for Community IT. Thank you for joining us for part three of our series of discussions on high tech and cutting-edge technology for nonprofits. 

Today, our three senior experts will be discussing Artificial Intelligence (AI) and Machine Based Learning and how these technologies are already impacting nonprofits. 

If you missed part one on virtual reality or part two on crypto and blockchain applications, you can find them if you subscribe wherever you listen to podcasts. I’ll let our experts introduce themselves and get right into the discussion of AI.

Johan Hammerstrom:  My name’s Johan Hammerstrom. I’m the CEO at Community IT. I’m happy to be here today with Steve. Steve, you want to introduce yourself?

Steve Longenecker:  Sure. I’m Steve Longenecker. I’m the Director of IT Consulting at Community IT.

Nura Aboki:  My name is Nura Aboki. I’m an IT Business Manager and Senior Consultant at Community IT.

Steve Longenecker:  Do you want to take a stab at artificial intelligence? What are your thoughts about that?

Johan Hammerstrom: Yeah, it’s interesting. I mean, it’s the kind of thing that I was pretty skeptical of for a long time. Just feeling like it was a lot of hype. I think part of the problem with technology, with high tech in general, is that there’s enormous amounts of money to be made. And the way new technology gets funded and promoted really lends itself to a lot of hype and exuberance that isn’t always warranted, especially with new technology. So it’s hard to know, is this real or is this just a venture capitalist trying to drum up interest in their latest project they’re funding. But it does seem like there’s things happening now in artificial intelligence that are real and legitimate and different. 


My understanding is that a lot of that is driven by two things that you need to make artificial intelligence work.

Most AI is using a neural network based approach where you build this neural network and then you train it much in the way that human intelligence is thought to work. You’re just showing patterns. In a reductive sense, human intelligence is pattern recognition, and responding — recognizing patterns and responding to patterns. And that’s a lot of what artificial intelligence is. You create a model and then you train that model, and the more data you can use to train the model, the more accurate the model is going to become when you start asking it to do more and more sophisticated things. And so, large scale models and large scale data sets are something that the modern world of the internet has really enabled.

And so large scale models are available now for the first time ever because of these enormous data centers that Facebook and Google and Amazon and Microsoft to an increasing extent have created. And those four companies have created data centers that are significantly larger than what any other entity on Earth has created. 

And in fact, when the US government wants to open an enormous data center, they contract it out to one of those four companies. So because of the scale and the size of the company, basically they’re creating these enormous computing environments that have the ability to create models at a very large scale in a way that has never been possible before. 

That’s one half of the equation. The other half of the equation is the data sets to train them on. 

In the same way these companies have amassed over the internet enormous data sets. 

Steve Longenecker:  Billions of pictures of dogs, zillions of pictures of cats. If you want to train a computer on why people think certain things; certain cat videos are funny and other cat videos are not. There’s a billion cat videos with data about how many of them are viewed, which ones are viewed more, and just feed it in.

Johan Hammerstrom: That’s exactly right. That’s what TikTok is to a large extent. It’s an algorithm that is proprietary, so none of us know how that algorithm works, but the algorithm is being trained supposedly on the behavior of millions and millions of people interacting with that product. It modifies and adapts over time. And that’s what all artificial intelligence models are doing. 

But just using Google search is using artificial intelligence. If you’ve uploaded your photos to Google Photos and you go in, it’s going to start surfacing things to you. Google Photos has become remarkably good at identifying one person over a long period of time. And I know this because I have kids and my kids have friends and their photos are in the Google Photos account. And it always amazes me how well Google can identify the same person over many, many, many years. You think that’s an aspect of human intelligence in some ways, to visually be able to identify, but machines are now gaining that capability. 

I think about some of the auto-completion. If you start writing an email, Outlook will automatically start finishing your sentences for you. And for a long time it was terrible. It seems to have gotten better. That may be like a boiling the frog situation where you don’t realize it, but one day you start to type an email and you realize, wow, Outlook is better at this than I am. That may be where we’re headed.

Steve Longenecker:  Yeah. It’s worth telling folks that are listening that right now, I don’t think they’re necessarily that they’re free, but I don’t think they’re expensive, you can get computer generated images. It’s essentially a website and you can say, I want a picture of a dog jumping over a fire hydrant. It’s not finding on the internet a preexisting picture of a dog jumping over a fire hydrant. It knows what the word “dog” means. The AI knows what the word dog means or what it has meant in the past, and it knows what a fire hydrant is.

So it gives you an almost photograph-like image of that. And then you could probably say, no, I wanted a small, yellow dog, not a Great Dane, or whatever. And then it would adjust all that.

Text is often the lead on these things because there’s so much text to train with. But you can get things to be written up based on a couple sentences and it finishes the paragraph for you. So those things exist today. They’re clearly imperfect. There’s oftentimes flaws with it, but it’s already the case that people who generate that kind of material professionally are feeling threatened by it.

And there’s a whole moral dimension to it. I wouldn’t call them scandals, but there’s been outcry that so and so was writing a blog and they used an image. Rather than pay a human being for an image they just went to one of these AI systems and got the image out of that and they paid their $300 a year membership fee for that service and unlimited images are available to them. 

Don’t quote me on any of the payment systems or pricing, but my impression is that it’s not that expensive and that this is something that you can do. 

Our experience is that in some ways the nonprofit community is, (we talked about leading edge) is like the trailing edge oftentimes, because when it’s leading, it’s very expensive and you wait for the market to make things more efficient and less expensive and that’s when nonprofits get interested. 

Things that humans do now, but are repetitive. like answering the phone, that’s the classic case. It’s not even so much artificial intelligence, but computers have more or less replaced auto attendants. But they’re going to get those auto attendants, we experience them all the time when we call technical support. You spend the first five minutes talking to a computer that wants to route you to the right person and ask you in your natural voice to say, can you tell me what this call is about? And then what you can say is “representative!” 

But if you just say, “I’m calling because I need help with my copier,” it knows. It can understand you, it seems nine times out 10. And it’s all amazing, and I think all that’s just going to get better. It seems impossible now that nonprofits would be taking donations via artificial intelligence, but at the same time, why not? That will happen. I don’t know when it will happen, but it will happen that you’ll call a number and a computer will take your donation and it will be a good experience, not an irritating one.

Johan Hammerstrom: Yeah. There’s already GitHub, which is the leading repository for code, for developers. A lot of people who are writing computer applications will store their code in GitHub, which is owned by Microsoft. GitHub developed an application called Copilot. If you’re a developer and you’re coding, you can run Copilot in a window. You’re coding in one window and running Copilot in another window. Copilot is watching what you’re doing and making recommendations to you about how to code what you’re trying to code. They’ve found that coders who use Copilot are something like 40% more effective and efficient than coders who don’t. It’s having a meaningful, positive impact, it’s making them more efficient, and it’s an area where they’re starting to see AI become more sophisticated in terms of the levels of code that it’s able to assist with.

Steve Longenecker:  Although my impression with that tool is it’s not particularly creative. It’s more helping with the kinds of things that are a nuisance to coders, like copying the same subroutine over and over again. I know that there’s that module and all I’m doing is putting an input in. I want the module to process it and then give me back the input and I have to look up the name of that module and then check it and make sure it’s the one I’m thinking of. If Copilot just tells you that’s the module you want, that just saved you X seconds or minutes. 

I don’t know if it would qualify in the neural networks kind of thing, but I am a faster composer of text when I’m typing. I’ve gotten much, much better about not backspacing when I mistype and just blazing along because half the time (it’s Microsoft Word), after I’ve typed it, will auto correct it. And the other half the time it just has a red line under it. 

And I can do a spell correct at the end, but I’m not like, oh no, that’s not spelled right, back, back, back type it again. I can see that, but I just plow through and the artificial intelligence saves the day at the end. 

It’s not like I don’t know how to spell and it’s not like I don’t know how to write. It’s kind of that difference between virtual reality and augmented reality in some ways. 

Now, who’s to say that we aren’t going to have AI completely replacing whole people at some point? I won’t be surprised if the receptionist job is gone. That may already be the case, largely. But I think a big role for AI in the nearer term is to help people who are doing jobs do them more easily.

Johan Hammerstrom: Yeah, I think that’s right. And at the risk of getting too philosophical, I believe that creativity is something that is uniquely human and there’s no point at which the machines that we build will have their own innate creativity. They can enhance and augment the creativity that we bring to the table, but I think it’s impossible for them to be creative on their own.

Steve Longenecker:  Yeah, we’ll have to wait and see on that. I don’t know if I agree or not.

Johan Hammerstrom: That’s angels dancing on the head of a pin, for now at least.


Are there any other interesting developments that you’re following in technology?

Nura Aboki:  Well, one neat solution I’ve seen has to do with the area of natural language processing, essentially chatbots. 

Some nonprofit organizations are adopting chatbots to be their virtual assistants in communicating with their volunteers, their members, and donors. That technology has really improved to human-like capabilities. It’s having a conversation, for instance, with the Google Lambda algorithm natural language processing solution. 

There’s potential that in conversation, whether it’s audio or in chat, it may be hard for another human being to distinguish whether it’s a real person or a machine or a computer. It’s an impressive technology that if done the right way can really help a lot of people. A lot of nonprofit organizations can expand their reach and really achieve their missions because it’s a partnership that would unlock so much potential for nonprofit organizations. That’s one aspect of advanced technology that I’ve seen more in the AI realm. It’s called natural language processing and specifically has to do with machine learning and deep learning algorithms.

Johan Hammerstrom: Is that technology that’s starting to become available in things like Microsoft?

Nura Aboki:  Yeah, I’m seeing that Microsoft has its own chatbot. Basically, the chatbot is built using AI or an artificial intelligence machine learning algorithm. So yes, that is becoming available in Microsoft. It’s becoming available in some Google products, sometimes providing analytics and data, deep insights to customers and clients.

And the technology itself, there’s been concern about some ethical use of it as it continues to evolve to be human-like. So as nonprofit organizations adopt these AI solutions, they’re really thinking hard considering the ethical issues related to having this machine learning tool make decisions on their behalf. So that is something that they have to tread carefully.

But yes, in terms of utilization of this technology, it’s available. Sometimes when you do a chat session with, let’s say, one of the large vendors, you’re not actually chatting with a human being.  That is one of those AIs that is responding to your comments and looking at what they’ve learnt by other reinforcement learning techniques to give you answers that may be suitable for you.

Johan Hammerstrom: One of the analogies I’ve heard is that it’s sort of like Clippy, but taken to the next level. Clippy, on a superficial level, was pretending to be AI, but in actuality, the things that Clippy suggested were never very helpful. And it was kind of annoying and it was corny and much divided. Clippy was taken out of Office. I think Office 2000 had Clippy, and then they got rid of Clippy.


Would you think that we would start to see more of that in office applications, like a digital assistant that would help you when you’re putting together a report or presentation?

Nura Aboki:  Yes. And there are digital assistants that will help you write a grant, a draft for a grant using these artificial intelligent solutions.

Johan Hammerstrom: Oh, wow.

Nura Aboki: And similarly, if you are in the US or you are on the internet, you may come across a solution called Grammarly, that reads and corrects your sentence. In effect, that’s just a basic  solution that can really add some input about the grant. And they may have already developed models about a variety of donors in the industry, and drafts and they will draft a copy of a grant that you may be interested in sharing. Doing that, they will help you edit it to completion before you send it out to your donors. 

That is a solution that is currently available actually for testing for nonprofit organizations that are interested in having their own AI that will help them draft grants. 

The other aspect of this, I am seeing an AI solution like this installed on a robot, and this particular robot would then help in cleaning the environment. So it would go into a trashcan and sort out the recyclable items from the non-recyclable items. In environments like universities, or spaces where we don’t really adhere to recycling best practices, this robot can actually go in overnight, sort this trash, then hand it over. 

Johan Hammerstrom: Wow, that’s amazing. That’s really cool.

Well, and you could imagine that you’re talking about physical trash, but when you look at some of the databases, I think of some of our data sources. I feel like it’d be great to have a robot go in and groom the data for us, clean it out a little bit. I tried doing that myself and ended up deleting my calendar accidentally, so.

Nura Aboki:  Oh, no!

Johan Hammerstrom: I could have used an AI system in that case. Well, that’s fascinating. Yeah. 

What’s the name, Nura, of the grant writing assistant that you were referring to? Do you remember? 

[Carolyn:  The website is fundwriter.ai]

Johan Hammerstrom: Well, that’s really interesting. I think the fear, obviously, is that it’s going to take jobs, right? The fear is that people who are writing grants for a living will end up losing their jobs to this artificial intelligent grant writing assistant. That’s the negative view. And there may be some truth to that. 

I think the positive view is the trash sorting example. It’s a much nicer one because that’s something that no one really wants to do and the idea is that these assistants would help create greater efficiencies in these sorts of tasks that can be very time consuming.

Nura Aboki:  Certainly, there is that concern about the loss of jobs, the “machines are taking over” type of feeling. I see that jobs are going to shift. We’re going to retool and rescale  the workforce.

Someone has to build this robot, someone has to code this AI. Even though the AI these days are calling themselves self-learning, reinforcement learning, but in the initial stages, there’s going to be a human being trying to develop the solutions. So one may need to rescale, retool and I agree there might be certain losses. But there will be a lot of gains and if we plan ahead, strategize, we will get ahead of whatever risks that may be presented and be able to resolve issues as they come.

Johan Hammerstrom: All right. I’m going to ask you a couple rapid fire questions here, Nura. I’m going to ask you to look into your crystal ball and just give me your best guess, okay? 

I remember back when the iPad came out in like 2011-2012, it was sort of the death knell for the laptop, right? iPads were going to replace laptops and it was the end of the Wintel Computer as we knew it. And here we are, 2022, we’re each on laptops recording this presentation. 


Five years from now, are people still going to be using laptops to do their work?

Nura Aboki:  Yes, I think people are going to still use laptops. In the enterprise, in the business environment, five years is still a short time for us to see a significant jump to tablets only, or some other form of being productive at work.

Johan Hammerstrom: What about 10 years from now? What do you think?

Nura Aboki: Oh, yes. So then 10 years from now, I sense there will be some other technology that will come and replace the laptop, itself. 

The idea of augmented reality, VR, those solutions would have been solidified 10 years from now. They will begin to replace the need for having a keyboard, mouse and a screen in front of you. You probably have a gadget somewhere strapped on your body that takes you into VR or AR.

Johan Hammerstrom: All right. 


What about email? Five years from now, will people still communicate with email?

Nura Aboki:  So we’ve seen a shift in using email and settings. People are using Slack, WhatsApp and other mobile applications. I think email will still continue, but it’s just going to be more on mobile. So having to use laptops, five years from now with so many people checking email, but there will be continued use of mobile devices to access email. And then there’ll be mobile apps like Slack and the other brands that are basically trying to replace the email entirely that will be adopted.

Johan Hammerstrom:


And 10 years from now?

Nura Aboki:  Well, we said that about email, many, many years ago. I think we are still going to see email in the space. There hasn’t been a solid potential replacement. We’ve seen experiments happening, but I think 10 years from now, we may still have email in our space.

Johan Hammerstrom Yeah. And the words of Mark Twain, its death has been greatly exaggerated.

All right. What about office networking? 


Will offices still manage their own WiFi systems or will everyone be on 6G everywhere you go, there’s a 6G, 7G network, you don’t need private office networking anymore five years from now?

Nura Aboki:  The latter, where you mentioned the 6G cellular advanced mobile connectivity, because the speeds are matching up or even exceeding current WiFi speeds. 

So if you have a 5G connection, you run a speed test, you’re approaching gigabit speed. It gives you flexibility, it’s mobile, it’s available virtually in and out of the office. So there may not be a consistent reason for wanting to keep building out WiFi networks. 

The only concern I have is just how do we make sure that they’re secure private networks, as part of the larger cellular network advancement? So that client can say, oh, this is my own portion of 5G and I get dedicated speeds. And I think that’s the likely evolution in the mobile space. It may be a limited technology right now, but in the future, we’re likely going to see that.

Johan Hammerstrom: Twitter, 


Will Twitter still be in business in five years?

Nura Aboki:  Well, it depends on the leader. The leader and the owners of Twitter will make that decision for us.

Johan Hammerstrom: What’s your guess?

Nura Aboki:  We still see in the next five years-

Johan Hammerstrom: If you had to put money on it.

Nura Aboki:  Well, the leader certainly has vision and I think they’re still going to be in business in five years, right now.

Johan Hammerstrom:


Will Apple finally figure out how to make the iPad a productive device that can replace someone’s laptop fully?

Nura Aboki:  I think they’ve already figured it out. I’m looking into my crystal ball and the R&D has figured it out; they’re just slow to roll it out. I think we’re not quite ready for that change. And perhaps as a business case, they don’t have a business reason to make the iPad replace the MacBook lineup.

Johan Hammerstrom:


Will Google Workspace make more significant inroads as a business productivity solution compared to Microsoft 365?  

I think they probably have like 15, 20% of the market right now. Will they increase their market share in that area, or will it remain roughly the same?

Nura Aboki:  In the business space, Microsoft is leading productivity solutions targeted at the enterprise. I think they will continue to make investments so they can take some more market share. But I don’t know if they will be able to exceed Microsoft’s strategic investments. They’ve all gotten us hooked. So it’s very difficult to break that cycle. 

But the next generation of users are using Google. It’s possible that they will gain some share, but I don’t see Microsoft falling behind

Johan Hammerstrom: All right. I was going to ask you whether Peacock and Paramount Plus will still have standalone streaming services in five years, but you don’t have to answer that.

Nura Aboki:  The experiment is going to be over.

Johan Hammerstrom:  I agree.

Steve Longenecker:  I do think that in 2012 there was this up and coming thing called social, and it really was well leveraged by a lot of sharp people in the nonprofit space. It was adopted by nonprofits more quickly and to good effect. 

There are counterexamples of where there are buzzy new things that by getting on board and being an early adopter, not too early, there’s no point having a Facebook page when you were the only nonprofit with a Facebook page, but not waiting too long. 

A lot of nonprofits got a lot of mileage and obviously now you’ve got to have it if you’re a nonprofit that faces the public. That’s a requirement if your mission is to be out there in front of the public, you need to have a social presence and someone doing that. So there are counterexamples. 

But I think you’re right. When it comes to blockchain, AI and AR; probably wait and see is the best advice for now.

Johan Hammerstrom: But to your point, being aware is also important and these things can change on a dime too. One day there was no iPhone, the next day there was an iPhone and within a few years, everything had changed. Same with social media, same with high-speed internet back in the early 2000s. So while certain technologies take a long time to come to fruition, when they do arrive-

Steve Longenecker:  The tipping point.

Johan Hammerstrom: It’s a tipping point, and it’s important to at least be aware of that. And if you’re a nonprofit, you probably have staff in your organization who are more connected to technology. Younger staff tend to be more connected to technology. It is a good reminder of the importance of technology leadership that is aware at least of what’s happening in the world of IT. 

It’s kind of a recurring theme when it comes to buzzy technology for nonprofits. Don’t get into it. Wait until eventually somebody will bring it to you. There was no point in anybody getting into really, really early versions of the internet just for the sake of having it if they weren’t actually using it for something productive. Eventually DSL came and you could get on the internet at an affordable price.

Steve Longenecker:  Right. I think that makes sense for the topics we’ve discussed so far. When blockchain is a compelling solution for your nonprofit, you’ll know it. In the meantime leave it alone, let other people sort it out.


I think that’s true currently for artificial intelligence and virtual reality, as well, especially for small nonprofits. But for nonprofits in general, maybe just beware of buzzy new things. Wait for the buzz to go away and for it to become a more mature product.

Carolyn Woodard:  Thank you for joining us for part three of this discussion of nonprofits and high tech. You can find parts one and two if you subscribe wherever you listen to podcasts, on YouTube, or on our website. Get in touch with us on our website [right here!] to suggest future topics you’d like us to talk about.