Alvar Soome - "How to eat a wooden carrot"

[music]

[music]

So hello everybody. Uh first uh a little bit of warning information as you um may notice that English is not my first language and actually it is not even my second language. It's my fourth language because in my primary school there was no teaching such capitalistic uh slang. That's it. Uh so may God have mercy on your ears and shing and is uh very easy topic. How to eat a wooden carrot? Normal logical question. Let's dive into it. U why such a question? Let's imagine scenario where uh our government is very worried about us and our food security and they want to secure some uh food for people in distressful times. So they announce big

tender and procure carrots. A lot of them everything fits. They are orange. They are right shape, right? But they are made of wood. And farmers are of course pissed. And how do you uh order such thing from shop that makes wood and carrots? They are not edible. And what uh what is the government then um planning? If you think that they are planning to order edible carrots, you are wrong. They are teaching you how to eat wooden carrot. Cut the to small pieces, grind it into your food, smother it with ketchup. Then maybe it's edible. Try. We are very worried about you. We did everything right. Look at the tender. Everything is by the law.

All requirements are met. What What are you talking about? Okay. Then the screams get very loud. They make new procurement and now there is a requirement that the carrots must be edible but with Bluetooth then it's okay. uh sounds uh a little bit uh not so good but and too topical maybe but we are doing it. Uh best example for that is uh 8 years ago uh uh Estonian government started uh international tender uh to procure national alert system. Every everyone who was in uh uh info technology or telecommunication sector they looked at the tender and asked what is this? This solution is already there. It's named cell broadcast. But you are doing your own thing with the blackjack

and hookers. And why? And of course government said that hush hush we know what we are doing. And they started build our own alerting system. As you remember it doesn't work as well as we need to uh it to work. My friends got messages about the drone hovering around when they are sleeping in Singapore. Veryformational, very good information. Uh and then when there was testing that government said, "We want to test how this uh alerting system works. Please turn your phones uh into into sound and ensure that they are not in silent mode because then you don't hear our our testing. Maybe it's not the right way to approach that. But yeah, luckily for us

they are very [clears throat] smart smart people and now they are saying that hooray next year we get the broadcast system. for the information. Even Latvia has it already two years. And if Latvia does something better and faster than we, then something is completely wrong. Uh maybe we look into that. Uh and why they did it, nobody knows. If if you ask why you order such such thing and actually in Dender there was such requirements that uh uh excluded cell broadcast you couldn't uh uh offer cell broadcast system because it doesn't meet the requirements that our government put in place which is stupid because it works all around the world. If you are are some frequent traveler, I think in

some places you have experience tele broadcasting alert. It uh works when your phone is turned off silent doesn't matter. It raises such alarm that even dead came alive. So uh this is good example and why why it happens this is always puzzled me why such such thing happen we uh took 8 years to find out that what we are building with millions doesn't work and how it could be because I'm pretty sure we have very smart people working in our our government uh buying ing things, ordering things, making tenders, procurements and how it happens. This is a always mystery for me especially when uh uh first 21 years I worked in private sector then I went to

the public public sector and I was uh how it is constantly amazed uh when there were [clears throat] funny uh proposals and not so funny uh uh um uh time uh calculations but uh I tried to explain how we are are made have made formula that results in disaster. I don't know how it happened but uh I think this is my personal opinion uh four years of uh research no no research first uh component of the formula is ignorant client sometimes we want something but we can't describe it correctly but We we want it. We need it. And uh nobody asks why why that way why such approach and why such requirements inhouse technical knowledge

has fed is very minimal. People know what they want but they can't describe it to the technical people. Second component greed the contractor. I know private sector doesn't want to hear that but come on if I look at the contracts and the money that is given to the private sector for IT development projects then this is not normal. uh last couple of weeks couple of examples as uh self-service environment website 3 years 4.8 8 million euros surf service website. Why? Then our our social social payment system guys uh I'm already confused this this guys second third fourth I don't know nobody knows uh five years 28 million euros this is I don't even know how to explain that nobody knows

and third component is easy money this is the most dangerous part Actually uh we we are given a lot of uh uh money from Euro European funds TEF and uh they come with requirements and one requirement is that mostly it is usable only in new developments and of course we want to spend that money. free money of course you want to spend it and we then build enormous IT projects innovation uh but somehow they always forget what comes after that if the project is ready first version then the safe money goes away and you can't uh uh use safe money to better things or host things or backup things you must find that money in more by that and it always comes with

big surprise that how we got here again and I are always I tried to explain uh in public sector that do you understand that project doesn't finish uh with that stage that the project is first version is ready project is finished when it's closed not when it's ready to use project is finished when It's closed. It's archived. It's deleted. And even this closing is not free because in some cases you must uh uh give our national archive a very special pack format, but it it's meant to be how to pack up database and uh uh business logic. very special format. But yeah, that's the reason because in-house money is always not enough. Always safe money is easy

money and easy money is risky money if you don't uh look to the future and make make arrangements that this is a viable project. There are many examples. There is uh uh our uh uh school managing system named Tafel. It was made with ch money. It is now in use but it's It doesn't work. Uh and there is no possibility to uh get their new functionality because there is no money. You can't do it with safe money. And then the result of course is disaster and no presentation is presentation where there is no catpicks. Where is my catpig? Thank you. So but uh this formula is very hard to solve actually and in AI century is even

even harder actually. My question always to the our big IT houses. Greetings Nortal Helix is you talk in big stages how you use AI in your development. Everything is easier, faster, better. And then I look at the contracts and ask it but it costs the same. How comes? Nobody knows. Uh but we must do something. And one thing that uh I think we must do is we must stop ordering working hours and bodies and start to order results. But it's very hard. It's very hard to describe the result that you want so everyone understands it correctly and you can uh after it's ready say that yes you delivered what I wanted. It's very hard but there is no other solution

actually ignorant client is yeah our own made problem and DC money is European Union made problem so if anyone has better solutions I'm big here and now I give some examples They are real examples a little bit anonymized but real examples. There was dropupal website and uh on my desk landed proposal for plug-in €30,000 everything it was calculated analyzed uh uh our developing partner was uh giving time estimates €30,000. Okay. €30,000. But what it does, what uh gets better? Then I ask some gazillion questions and I found out that it's very good thing. It saves time for one person work. four hours good yearly. How? And then I stopped talking. I just looked and they walked out to the room

and I didn't hear about it no more. And again my how I looked that if Luke's can kill then this is the one. Another example there is it doesn't even matter what it is there is collection of objects or objects are have object card where is all the data about it where it is what it is and there is some detailed data but the problem actual problem uh when I was working with some object I wanted to send uh this uh object link to someone to look over. It was not very um comfortable. It was hard. Not hard but not not logical. You must you were forced to uh uh push some button that open new

window. There was link that you must copy paste and and we we send request uh to developing partner then take take that link leave it there also take that link and put to the object card and thank you even don't develop that uh copy to clipboard no problem right mouse click and and we we we can manage And then came estimate. How many hours this takes? What do you think? What was the estimate to move that not? Not even move. Put that link also here and your quest correctly. 47 hours. I was um excuse me. Uh, how much? What do you do with 47 hours there? You go to the spa or what? But they were very gracious.

Uh, they thought about it and told, "Sorry, we understood wrong. It takes only 27 hours." But it was win 20 hours less. But actually it was just copy pasting one HTML tag to another place and literally it was that and uh our public sector is full of such small developments and big developments and uh we are know in the in the we are reached the situation when we Don't even ask the question why selfserving website takes 3 years and 4.8 million euros. Oh, it's important thing it must cost that much. No, maybe ask why is it even necessary? This is also very hard uh uh now thought uh to um explain to the public sector

that not everything must be digitalized. No clear example was this true paladon. This is was not needed of digitalization. No. So maybe here also if you hear something uh weird ask questions because best question is why why we do that what it solves what uh what is better after that and uh I'm always uh when my are in conferences conferences seminars then there are people that are talking about oh we now make 3 million such innovative thing and 5 million such innovative thing and when I ask that maybe for a moment let's stop making innovative things and a little bit make working things that are needed then confusion is big no innovation a little bit

okay I know I'm bad person person. And uh what is another reason how we are here? We actually painted ourself into a corner in Estonia's Estonian saying is is from uh navigator bricks movie. uh actually like always with with regulations and laws that's the problem uh when I started to working at public sector then first thing that was very painful for me my brain doesn't work that way uh my brain is round and that thing is square it doesn't work is our uh [clears throat] national tender and procurement door. Uh this is amazing. Uh when I need to buy some I have money, I have gold. I have accept accepted my board has accepted that that is needed

and I can't go and buy it. I must announce tender to buy some box. And then it takes three weeks and 20 pages of documents and all people are working like mad and when I ask why it is so then nobody knows. Uh but actually there is a problem. I feel almost every day that uh this makes my work harder. It's painful. It's very painful when I heard procurement law I get shivers. Uh but then Andres asked me that okay there is problem but how how can we solve it? What is the problem? Point at it I can't. My answer is everything. Something went somewhere very wrong and nobody knows why. They wanted I'm I'm pretty sure they

wanted only good things. Nobody wanted some harm or let's make it harder to order something and let's make it pricier. No, but uh because it is such a legal maze then we give the edge to the big IT companies. Again, greetings Nortal helmets and so on. talle my favorite uh because they have the uh comp competence and experience how to handle the things and how to explain things and how to uh play with the system. They are very clever. Uh I don't like that. I know I'm bad person and I ask ask questions. They don't like that. But somewhere something must change. We can't go on that way that we make social payment system 5 years for 28 million

euros. This is not normal in a AI century. You give it 28 million. No this maximum amount of money and five years. We don't know what happens after 3 months and to plan plan ahead five years. You commit unreasonably and this is bad. And when I ask why you do that why why do don't you split it in smaller pieces manager pieces year is already very long time then answer is nobody knows. So we have we have we all know that we have problem but the solution is hard and actually if I honest I'm I'm pretty I'm pretty straightforward Kai then I see the problem but I still can't see clear solution. That's that is my problem

but we are talking about it with much smarter people than me. Let's see if he if he go somewhere with that. We have some ideas but all all those ideas have bad sides. They they sol solve something but they create more new problems. So it's hard a everywhere everyone wants to talk about a artificial ignorance correct there is uh same problem as uh with digitalation you you don't need AI everywhere this is very painful to uh for I enthusiast and I educ educators but you don't need AI everywhere. I know people who have no use of AI in their uh field of work and if you want to know who those people are come and

ask me afterwards. Those are not one or two people. I have many who even uh close their websites because they don't need web also. They have no use for that. And one problem with this AI and AI narrative is first thing where everything goes wrong is that there is a like this narrative that there is one AI user who is waiting for someone to give him right tools. There is no one common AI user. We are very different needs. If I would have development company, of course I use AI. Uh but in office work, how where people are very have very different needs. But at the moment uh uh government HD AI uh program is forcing

everyone that learn AI learn AI tell or you lose lose something I what do you lose I don't understand people are reading reading about it hearing about it and they are asking it but AI gives solution to to everything and riches and love and happiness I don't have that. I want that. How can I have that? Then they go to the strange trainings where strange things are taught and they went go back to the work and say to the IT guy that I want install everything and admin rights and the IT guy says no. Uh this is uh hard. uh and uh we are working uh to solve that problem that we have AI tools for

every every need but sometimes I reached the conclusion that AI is not needed here couple weeks ago one person came to me that I want better AI and okay why what is the result of better AI I didn't even ask what is better AI optimizing workflow. Huh? You don't know. Nobody knows. Uh but yeah, using AI is easy. Uh but what can we do? Actually when you are uh have been in IT sector 20 30 years I have been 34 years uh uh in IT using IT no something like that 26 years in uh uh IT work field then uh there is actually nothing new here we have seen it many times some new things come new thing

comes They promise everything. Paradise, nirvana, happiness. It solves everything. Then uh it turns out it doesn't solves solve everything. It solves some things and it creates some problems. Uh but now everything is on steroids. That's the problem. It's multiple times the usual problems and problems are everywhere and we need to manage it and there's always first people to benefit from new technology is surprise surprise bad people of course and we are running running behind trying to fight them. It's hard. They are quicker. Maybe maybe not quicker or smarter, but they don't care. We are bound with laws, maybe even morality. They don't have that. They don't care. They only want your money, your credentials, your information. That's it.

They don't choose your their weapons. We have limited possibilities to defend ourself. There is the big problem. But main thing we I was in Taltech and we discussed that uh what we must teach in those times to the students uh that when they leave the university they have the necessary uh knowledge and uh we agreed that one thing that you don't need to especially teach at the university is actually AI It's everywhere. It's a tool. It's used everywhere. Uh but uh you must teach uh logical thinking, time managing, functional reading, algorithmic thinking. Those are the problems like my wife is teacher and he told me that no Alvare there is not not anymore there is problem in

functional reading. Now there is reading problem. They just don't read anymore. And this is very bad problem. So what can we do? We must use your brains. You must use your thinking thingy. Uh when you use AI that the most harmful thing you can do is use it as copy based tool. I see it everywhere. No thought inside. Good enough. And now you you do that. And uh I'm starting soon to ask that to send that text to me. Sent the prompt. Then I understand what what what you needed because this wall of text is amazing. Four pages. Nothing. not one logical thing inside but oh boy it's beautiful and don't do that and of course

we are adults we are we have already our paces we are our pace knowledge most harmful is AI in schools in teaching if you don't use the AI to learn but to do then This is bad. This is a cognital downfall. Of course, we are are already again in the front lines. We are making our own educational AI tool with blackjack and hookers named it. uh and the idea behind this ity is this is AI that doesn't give answers but it pushes you to the answer and when I heard about it I I had only one question have you ever seen a child of course they turn to the model that gives the answer and and of surprise

surprise that happened uh and actually it it works a little bit uh but uh at the moment statistics so final statistics are coming out in December but at the moment the feeling in the numbers is yes there are many users some 5,000 students are using the city but the problem is that the students that are using it were not the students who had problems with learning. They had no problems and they are using it to learn like it was made. But students that had problems, they are using another AI that gives the answer and the gap between those children is growing and growing and this is bad. And of course we have uh [clears throat] very good ideas about

uh how to give our our AI agents personal codes. This is my absolute favorite. It's I want my Excel also to have personal code. Thank you. Uh this this idea is fantastic and it it's uh again on the table. It died couple of weeks ago that no no we don't give personal codes to AI agents you understood us wrong and now yesterday is initiated that yeah we do I don't understand this is stupid so yeah how to then conclude everything uh situation is bad why nobody knows uh There are many problems. Problems are with our uh uh with our public sector. We don't have too many good technical people there. Then that uh so they can

make good contracts and explain what we want. We have greedy contractors. Don't start with that. This is pure greed. uh when they are crying at why you are um in the public sector why are you doing in-house development to give the money to us no of course they want the money to them but uh the problem is not that public sector wants to take the money money away or work away from them problem is that some people understood that something is wrong because selfservice website doesn't cost €4.8 million. The problem is also there problems are laws regulations and problem is overusing of AI. So we have very complex bunch of problems and there is very hard to point to one

thing that the problem is only there let's solve that there is no such thing this complex problem and as we all know in it where there is complex problem and you try to solve all the problems in the same time then you make new problems That's the logic behind it. So, use your brain, be reasonable, be curious, ask questions. Most important question is why? And that's it. uh this is how AI analyze a situation and make suggestions and this is the logic I don't understand that when we started to brag about that my AI uh breached something and this is cool thing I don't understand that and my AI will kill more people and faster

that that is better AI Oh uh but I'm very stupid. So if somebody knows how it's possible, tell me afterwards. And if somebody wants to uh hear about more uh [clears throat] intriguing examples of uh government develop IT developments, you can also ask me. I have a lot of them to those were easy ones I showed you. So sorry about that. I ruined your morning. Thank you very much, Alvar. Now, we do also have some time for questions. So, does does anybody have questions? We have a a throwable microphone and one in the back as well. So, uh if you have questions, uh raise your hand and we'll get a microphone to you. Um everybody

still waking up in the morning. Yeah, I have a question in the front. Hello. Uh yeah, thanks for the awesome interesting presentation. I had some feeling that this is how things work in private sorry uh public sector and now I'm I'm more afraid that that's even worse that I thought how they are. Uh so my question is like the problem the 47 hour problem that you got for 27 hours. So did you like get an explanation what they actually do for that 27 hours or or that was like just they copy pasting the HTML tag and that's that. >> Oh yeah the expl explanation was beautiful and st and stupid. Uh but they gave the explanation and actually

this is also part of the problem. We needed that and we had the money money and I had no more um energy for that. I I took the win >> but actually it uh it should have been one hour. But is there like possibility to have this kind of model that like uh uh public sectors would have like at least one developer uh hired like full-time and >> we had >> and they didn't they were not able to do that. Uh yeah uh this was not the in-house development project then you can't you lose the warranty and you make more problems but we have in-house developers everywhere in MTA Smith. So yeah. >> Okay. Thanks.

>> Thank you. Any other questions? Um I have no hands yet. I'll I'll ask one of my own as well because um I noticed like you had the three different problems uh for the uh sort of why things fail. So you had the ignorance, you had the greed and the easy money, right? Um and solving all of them together is a problem as I get because you get more problems like how would you start solving it? Is it enough to solve one of those to start with and which one would be the first that you would have to solve? Uh easy money problem is the such problem that don't solve that. We need the money

uh but maybe solve the problem that uh you can use safe money also to uh uh improve your product and keep it running maybe. uh but the first thing is um ignorant uh client this is the first thing I would like to start solving and we have some ideas there but uh this needs very big decisions this is hard one because uh greediness is also something you can't solve tear you can solve greediness in ignorance not in greediness >> I get that but with the money And is it something like it should be it becomes sort of a movement that you have to start like hey can we also use this money for for improving uh stuff. So is

is it something that I don't know where would it come from? We would have to be the public sector asking Europe or somewhere like hey >> yeah yeah asking Europe actually we can we can bend the rules we we are famous about that that we overdo everything also we overdo regulations we make regulations that actually is not demanded by EU but we want to be good in front better than everyone else also in regulations. >> Okay. So, look at home first and then then Yes. Okay. >> Uh any other questions that have arisen somewhere? We do have microphones for everybody. >> Well, not not like you have to share but >> yeah, you can ask also privately.

>> Yes. Yes, you can you can find him around as well. So, thank you very much. Thank you. Thank you so much.