Learning Library

← Back to Library

Developers Debate AI's Real Intelligence

Key Points

  • Developers see generative AI more as a helpful “librarian” that retrieves and assembles information rather than a truly intelligent system.
  • JJ emphasizes that current AI lacks logic or reasoning, operating like predictive‑text by selecting the next most likely word from large datasets.
  • Because it can query natural‑language inputs against a company’s documentation, AI is best used as a specialized tool for accessing and summarizing existing knowledge bases.
  • The term “artificial intelligence” is considered overloaded and misleading; calling it “AI” creates confusion about its actual capabilities.
  • While the broader conversation hinted at concerns about AI taking jobs, the focus remains on positioning AI as an augmenting tool rather than a replacement for developers.

Full Transcript

# Developers Debate AI's Real Intelligence **Source:** [https://www.youtube.com/watch?v=MmgdcWA3bcs](https://www.youtube.com/watch?v=MmgdcWA3bcs) **Duration:** 00:15:22 ## Summary - Developers see generative AI more as a helpful “librarian” that retrieves and assembles information rather than a truly intelligent system. - JJ emphasizes that current AI lacks logic or reasoning, operating like predictive‑text by selecting the next most likely word from large datasets. - Because it can query natural‑language inputs against a company’s documentation, AI is best used as a specialized tool for accessing and summarizing existing knowledge bases. - The term “artificial intelligence” is considered overloaded and misleading; calling it “AI” creates confusion about its actual capabilities. - While the broader conversation hinted at concerns about AI taking jobs, the focus remains on positioning AI as an augmenting tool rather than a replacement for developers. ## Sections - [00:00:00](https://www.youtube.com/watch?v=MmgdcWA3bcs&t=0s) **Developers Debate AI’s True Role** - In a conversation with IBM’s developer advocate JJ Asghar, the discussion reveals that many developers see generative AI as a useful, but limited “librarian‑type” tool rather than a truly intelligent system, and they grapple with its impact on their work rather than fearing it will replace them. - [00:03:17](https://www.youtube.com/watch?v=MmgdcWA3bcs&t=197s) **Navigating Generative AI Libraries** - The speaker likens generative AI to an overwhelming library, explaining how to choose and evaluate the ever‑growing array of models while emphasizing the importance of sovereign, locally‑run solutions like IBM Granite to keep proprietary data secure. - [00:06:27](https://www.youtube.com/watch?v=MmgdcWA3bcs&t=387s) **Open Source vs Closed Source** - The speaker explains open‑source development, contrasts it with proprietary “cathedral” models, and argues that multiple eyes improve security, citing recent incidents. - [00:09:31](https://www.youtube.com/watch?v=MmgdcWA3bcs&t=571s) **Developers Prioritize AI Ethics & Transparency** - The speaker stresses that developers value ethical governance and transparent AI models—exemplified by the Granite initiative—while reassuring listeners that AI will not replace their jobs. - [00:12:34](https://www.youtube.com/watch?v=MmgdcWA3bcs&t=754s) **Developers: Embrace AI Beyond Kubernetes** - After thanking a non‑technical audience member, the speaker urges modern developers to prioritize learning and thoughtfully integrating AI tools—rather than merely slapping AI onto existing cloud‑native solutions. ## Full Transcript
0:00It seems like folks treat technology development as if there is an easy button. 0:04You know, just press it and it's all good. 0:06So I have to wonder, how do the people who are actually doing the work 0:10developing these programs themselves, how do they feel about AI? 0:14Or, the $10 million question, do developers of AI 0:18worry about AI taking their jobs? 0:21You know we want to know. 0:22So my guest today is JJ Asghar, developer advocate at IBM. 0:26And he’s going to take us inside. 0:28JJ, welcome. Hey, thank you so much for having me. 0:30First up, you're coming from the developer POV. 0:34So can you tell us from where you stand or where you sit, how are developers 0:38thinking about generative AI as a new development tool? 0:42The core problem with AI for developer standpoint 0:45is, it's not very smart. 0:48It's, it's not and it's not really, so you'll notice quickly that I will not use 0:54the term artificial intelligence because it's not intelligent. 0:57AI is an overloaded term now 1:00that has caused confusion in the market. 1:03There are certain tools for certain jobs. 1:05And the best part about AI is it is one of the best librarians 1:09you'll ever have in your life. 1:10Hold on. 1:10Is that a controversial take right there, JJ You know, as somebody who lives 1:14and breathes this every single day, I'm trying to tell you the truth here. 1:19Now, I want to spend a little bit of time on that then, 1:22because you really made a point 1:24to, you know, take away the intelligence part of it. 1:27Why can this not be considered intelligent? 1:30Oh, that's a philosophical question there my friend. 1:33But in short, what it is, is a is a program 1:37that is looking for the best possible answer to what you are asking. 1:43There's no logic inside of it. 1:44There's no reasoning inside of it. 1:46What it is doing is looks in a database and it says you're looking for apples. 1:51Okay, cool. 1:52Well, these words are really close to the word apple. 1:55So maybe you're asking about Granny Smith apples. 1:57Maybe you're asking about red apples. Maybe you're asking about green apples. 2:00So it generates to that question 2:03where adds to those words as that sentence. 2:07If you’ve ever noticed as you use generative AI, 2:10it comes out in word per word because it's actually looking for that next word. 2:14It doesn't figure everything out and then dump it out. 2:17It's like predictive text on your phone, if that makes sense. 2:20No, that makes complete sense. 2:21So then I'm wondering from your perspective as a developer, then 2:25how do you best go about using Gen AI to its maximum capacity? 2:30That's a great question. 2:31Again, it's another tool, right, where every single company 2:35out there, every single person, frankly, has a bunch of documentation, right? 2:39They have information that they need to store in, 2:43what some people call the second brain. 2:44Right. You know, 2:45you take notes or those notebooks, people have with notes and whatever. 2:48If you look at generative AI from that lens, where it is now a thing 2:52that you can query in natural human language or natural language 2:56processing to be able to ask it questions about those documents. 3:00Hence, the librarian, that becomes really powerful 3:04because now a company can have all of its documents 3:07and then instead of going to the HR representative to talk about, 3:11you know, your insurance policies from 1963 or, 3:15I don't know, whatever number you're thinking, right? 3:17But, instead of asking those questions and having them go look for this, 3:21now you have this thing that already knows about all that, 3:23all that data, or at least gets really close to that data. 3:26Well, then, since you started this, I'm going to go back and forth with you 3:29with this as a library metaphor, because I love a good library. 3:34But a library can also be super duper intimidating, like my university. 3:39You know, 3:39we had two levels 3:40and levels in the stacks, and so much information is housed within there. 3:43So if we're looking at gen AI in that way, there are just constantly 3:49and ever expanding options that are out there. 3:52How does one even begin to navigate and evaluate these different tools 3:56that are there? 3:56JJ, how do you know which book to pull first? 3:58Yeah, great. Wonderful question. 4:01One of the biggest challenges is that every single model, which is 4:05what is the brain, right, of the generative AI, 4:09every single one is there's different ways of designing and programing those. 4:13One of the best parts is, is there are generative 4:15AI models out there, just like something called Granite from IBM 4:18that you can run locally inside your own data center 4:22or inside your own country, which allows you to have sovereign AI. 4:27One of the biggest problems is that you don't want to as a company, 4:31you don't want to send your data out to the San Francisco Bay area 4:34and have them crunch the numbers and come back right 4:37that you're sending it across the internet. 4:38Would you send your secret sauce across the internet? 4:41No, that's a horrible idea, right? 4:43Yeah. 4:44So that’s 4:44the power of having these different models and different ways they’re designed. 4:49And the foundational model from IBM called Granite, it's a model 4:53that is designed to be able 4:54to run in your own data center and now you can train it, or what 4:57we call fine tuning to give it more skills and more abilities. 5:01Well then hold on. 5:02So you mentioned Granite. 5:03What you’re talking about, it sounds as though Granite is open source, right? Yes. 5:07You can actually get the paper, believe it or not, on my browser right now, 5:11I have a link to the actual paper right above your head, which is really funny. 5:17because I read it all the time. 5:18It is, it's a it's a math paper, so it is actually kind of hard to read, 5:21to be honest with you. 5:22But I do actually, it is really there. 5:25But yes, you can actually see exactly what IBM used to build the data set. 5:29So if we're using the analogy that, a model is a program, 5:33think of the data set as the source code, okay. 5:37That's not 100% true, right. 5:39People are going to pick at me because I said that. 5:42But if you're trying to keep that analogy in your head to understand 5:45the power of this, data sets are the source code for the models, so that 5:49that's actually what builds the models, it gives it the initial knowledge. 5:54As you know, I didn't say intelligence. 5:55Initial knowledge of what it has to understand. 5:59And then you put your knowledge or what we call fine tuning on top of it, 6:03of your company's PDFs or documents or whatever inside of it. 6:06Gotcha. 6:07Well, sidebar, you're going to start having me say, AK instead of AI. 6:11Now you're messing with me, JJ. I love it. 6:16Now, I love that you 6:17broke down Granite for us there, but in general, can you let me know 6:21a little bit more about why open source could be considered as a beneficial thing? 6:27And is open source always the ideal? So 6:31you, my friend are a philosopher deep down inside, I'm starting to get this. 6:34I'm actually an open source engineer right. 6:36So what does that mean? 6:38That means most, if not all, my code is out in the public, 6:41right where you can actually see the tooling 6:44and the work that I'm doing where you can literally just find me 6:47on the internet and be like, oh, this is what JJ is working on now. 6:50Right. 6:51That is the core of open source, where you're using what they call 6:54I think it's, Eric Schmidt wrote an essay called The Cathedral and the Bazaar. 6:59There's a cathedral 6:59where it's very top down mandated, which is closed source software. 7:02And then there's the bazaar, which is like, 7:05you know, like a marketplace where everybody's kind of working back and forth 7:09and you leverage all these engineers across the planet to make stuff. 7:13So what does that mean? 7:15Open source allows you to have multiple eyes on problems looking for stuff, right? 7:20There's certain security issues 7:22that have recently happened on that have hit the news. 7:25That one was a closed source system that caused a lot of people 7:29a lot of problems traveling. 7:30And then there's another one that was actually even worse, but 7:33on the open source side, but was caught before any major issue happened. 7:38And it was because some, some nerd out there couldn't 7:41actually access their server as quickly as they usually got, 7:45which was really, really interesting. 7:46So we had one that took down travel, which was a closed source system, 7:51and then we had another one who was just one nerd 7:54who was like, I couldn't log into my server fast enough. 7:57Oh, there's a backdoor in open ssh. 7:59This isn't good. 8:00And then he found the CVE, and he figured it out, 8:03and then put it out to the world and fixed it before anything happened. 8:06Okay. See, hold on. 8:08That's actually really counterintuitive to me, because I would have thought 8:11that the closed system would have been safer than open. 8:15Because when I think of open, I think, okay, people can just come on in here 8:18like, bad actors can come and do their thing and mess around with it. 8:22But you're saying that in this case, the open source system, 8:25because it was able to draw upon, experiences from people 8:30that weren't just inside of that actually ended up being a stronger force. 8:33Exactly. 8:34It's 100% that because you have so many more eyes, so much more experience, right? 8:38I mean, what is the whole story of like, why do you need different people in 8:42the room is you need 8:43diversity and the ability for people to come with different viewpoints. 8:47What is open source? 8:48But the, the way the nerds are doing, a true diversity 8:52where you have people who are who’ve been in the military, you have people 8:55who have only ever, who failed out of university, 8:58you have people who didn't go to university, 9:00all looking at the problem in different ways, and they all resolve it. 9:03And there's a handshake agreement 9:06inside those rooms that allows you to say, okay, this is a good patch. 9:09Let's go ahead and submit this so this fixes the problem. 9:13So you mentioned this a little bit before too, JJ, 9:16I believe you said the word transparency. 9:18So if possible 9:19I want to time travel a little bit and go back there and dig going into it 9:22because you know transparency, ethics, governance, these are huge questions 9:28when it comes to AI or AK, in your situation. 9:32So what really matters to developers when we're thinking about those big questions, 9:36when we are thinking about ethics and data transparency and governance? 9:39So, frankly, as a developer, somebody who gets to play around 9:43in the plumbing, not the porcelain but the plumbing of the world, right. 9:47The ethics and the and the and the governance of it 9:50is insanely important to me, because I need to know 9:53the thing that I'm working on, frankly, like, I'm a human. 9:57I like people. 9:58I don’t want to kill people. 10:00Right? Like, that's not something I want to do. 10:03What a relief. Yes. Yeah. Yeah, exactly. 10:05But, you know, if we take AI the wrong way, it can really hurt society, right? 10:10It really can. 10:11And having that governance, having that transparency in it, we can be the rebels. 10:16And that's what we're doing here with the Granite model and the transparency. 10:19So we're giving you an opportunity to actually see into 10:23how these models are made, so you can make good choices 10:26for your business and hopefully society as a whole. 10:28Let me take you back then to this idea that there are legitimate concerns 10:34for you to have when it comes to AI, especially as a developer. 10:37So I'm going to get really personal with you for a second. 10:40Are you and your fellow developers, are you concerned about AI taking your jobs? 10:45No, not at all. 10:47Not at all. 10:48There's some great stories around using AI to build 10:51software, and people are like, oh, well, why won't you just get the AI to do it? 10:55People don't realize until much later on in their career, 10:58especially as a whole, or they don't teach you this 11:00in university, if you go down the computer science and engineering, space, 11:06they assume that engineering is math. 11:09And a lot of like, you know, sitting there 11:12thinking abstractly to figure out problems. 11:16But believe it or not, software engineering as a whole 11:18is actually knowledge work, right? 11:20It's actually artistic also, 11:21where you have to think of problems and unique ways to do it. 11:24And back to the intelligence statement earlier, 11:27AI doesn't have intelligence, doesn't have logic to figure it out. 11:30It can regurgitate code that it knows about. 11:33But if I put my business and ask it to create something for it, 11:37and then something went pear shaped inside of it, I would have to have 11:39an army of engineers to unwind what it did to make it happen. 11:44And this goes back to the analogy of the librarian, where there are 11:47some code completion systems out there, including Watson code assistant, 11:52which is from IBM, that allows you to use it as a reference, right? 11:57Where you can ask it like it’ll give you suggestions 12:00to put in like if/then statement or stuff like that. 12:03As a whole, you would never ask it to build me a piece of software. 12:07You'd use it as a pair programmer, a programmer sitting beside you 12:11so you’re like, so I'm trying to do this, and you write it out as a sentence, 12:15and then it gives you a suggestion, and then you look at that suggestion 12:18and then you edit it to actually do what you're looking for. 12:22Right. 12:22It gives you kind of a framework, if you will, or a straw man of the problem 12:27that you're trying to resolve and then come out with that. 12:30Does that make sense? 12:31No. that does. That does make sense. 12:32And thank you for breaking it down in that way. 12:34I just have to give you additional props right now because as someone who's 12:37not a developer, you're actually making this make sense to me. 12:41And I just appreciate you for doing that. 12:42But now I want to give you a chance to actually speak directly to some of the 12:46developers that may be listening, that hopefully are listening to this right now. 12:50If you can encourage developers to do one thing as they move on 12:54with evaluating tools and building solutions, what would that one thing be? 12:59As a developer, and hopefully it's a modern day 13:01developer, you're looking at me right now. 13:03You probably spent some time in the cloud native ecosystem, right 13:06where we use this thing called Kubernetes, and we're trying to do all these, like 13:09these VM to Kubern pod conversions and all that jazz. 13:13We thought that was hard 13:14and we thought we were going to make a lot of money doing that, 13:16because that was going to be the next generation. 13:17Well, turns out, AI is two generations ahead of that and it's even harder. 13:22So what you've got to do 13:23is you've got to go learn this stuff, and this stuff is confusing as all hell. 13:28I'm not going to lie. 13:29And it is a completely different way of looking at it. 13:32But it's not just PhDs and Jupyter notebooks anymore. 13:36There's actual tooling to get something useful out of it, 13:39but you're going to have to talk to your bosses to understand that 13:43as much as the VCs of the world want you to just slap AI on the side 13:46of your company or whatever to say that you're doing it, there's a lot more there, 13:49and you will quickly realize that there's a lot to learn, and the best thing to do 13:54is start from the ground zero and learn what a token is. 13:58And as soon as you understand what a token is, 14:00then find out the next thing you need to learn. 14:01I want to invite you real quick to let me know, 14:04is there anything that you would love to share that I didn't ask you about today? 14:08Oh, actually, yes. Back to the open source story. 14:10So we talked about the Granite model and we talked about how all that works. 14:14Well, there's another open source project 14:16out there called Instruct Lab that is came out of IBM research 14:19and has been donated to, Red Hat that runs it now. 14:23It is basically that fine tuning narrative that we were talking about 14:27by putting your your company's knowledge or your knowledge on top of something 14:30like Granite to be able to do something. 14:32It's in its infancy of a project, but we really do need developers 14:35to come into our space to start helping us in here, 14:38because the more we have there, the more transparency we show 14:41and the more the ability for the things that I was talking about earlier, 14:45it all boils down to what we're trying to do inside of Instruct Lab. 14:52And there’s enough to learn here that will teach you 14:54the AI ecosystem as you’re going down this path. 14:55So you’ll be able to understand the value of this space. 14:57Developers, you hear that, right? 14:58You've now got your mission. 14:59You got your charge, JJ needs you. 15:01Well look, JJ, thank you so much. 15:03This episode has been hugely informative. 15:06And again, if you are a developer who's been listening, first off, 15:09thank you for being here. 15:10But I know that you're also going to walk away with some great intel. 15:13So once again, appreciate you, JJ. 15:16And that's it for today's episode. 15:17But y'all, please stay tuned for more because you know that it's on the way. 15:21We'll see you then.