Miss Aempyr

Climb bed 😂

Ministry of Health website has also been hacked. What is really going on? Site: moh.gov.gh

@gwapgotme @mrzeal7 @koboateng @The_Ofosuhene I am not a bro 😔

@theandy_Ke @mrzeal7 @koboateng @The_Ofosuhene Bro you don’t get what I’m saying. We know how to build the system that does it. I’m saying it makes 0 sense for ANY business to do. Our margins aren’t high enough to justify research and development into an AI system for any region in Ghana.

We lack AI centers as a country. That’s why we’re suffering

@mrzeal7 @gwapgotme @koboateng @The_Ofosuhene Yes, even if it’s not perfect, it will definitely be improved. And for Sam George, they have realized how Ai is about to lead the world and they are finding ways to reap from where they didn’t sow. My argument was solely on possibility not cost

@theandy_Ke @gwapgotme @koboateng @The_Ofosuhene If such a tech exists then fine, but it doesn't. Maybe in future, who knows. And how do you plan to bring it to fruition with Sam George breathing down your neck?

@gwapgotme @mrzeal7 @koboateng @The_Ofosuhene I understand from your economical point of view. as an engineer, you should factor in cost. My argument wasn’t about cost but possibility of the design, every engineer will definitely go in for a less cost design. almost every Ai is till supervised bcos they are not 💯 accurate

@theandy_Ke @mrzeal7 @koboateng @The_Ofosuhene He’s right. It’s going to be largely supervised. Which still means you need workers. It’s funny cause people who handle simple projects think they’ve figured out some key issue . But .. if you actually work in industry you know almost every major company heavily uses ai.

@mrzeal7 @koboateng @The_Ofosuhene But self driving cars are already deployed into our environment. People use it in Ghana here. I even mentioned that, you need to use Ai with other traditional methods till we get a machine that’s highly accurate. Ai is never independent, it still needs human supervision

@theandy_Ke @koboateng @The_Ofosuhene Tesla's self driving capabilities is massively overblown though. You'll not find a single Tesla without human supervisors watching over it like a hawk, contrary to what Elon Musk would have you believe when he needs to pump up Tesla's stocks

@mrzeal7 @gwapgotme @koboateng @The_Ofosuhene Exactly, as an engineer. Never think something is impossible, it may not function well initially, but there is always a way out. Even if it will take 100 years. Someone will improve your design

@theandy_Ke @gwapgotme @koboateng @The_Ofosuhene Well it's good to be optimistic at least, nothing's ever done by saying it can never be done. Our tax money though is not to be wasted where companies with much deeper pockets have tried and failed so far

@gwapgotme @mrzeal7 @koboateng @The_Ofosuhene Okay. As an engineer, cost is one thing that’s very essential in design. But you can’t downplay safety and cost future cost. Will rather people everytime to clean it or you will build a system that will do it everytime without paying. technology is to make things easy

@theandy_Ke @mrzeal7 @koboateng @The_Ofosuhene Do not play the tech game with me. Im trained as an electronic information systems engineer and a mathematician. I could hire people for a few 100 cedis and it would cost significantly less than what it would cost me to build an ai system that gets this done.

@gwapgotme @mrzeal7 @koboateng @The_Ofosuhene Yes, in the world we are living right now. Almost everything can be solved with tech. Just that it sometimes have limitations but that doesn’t mean it’s not possible. So to you, your limitation is cost but doesn’t disprove the fact that it can be solved with Ai

@theandy_Ke @mrzeal7 @koboateng @The_Ofosuhene This reminds me of deming’s quote where he says people have this sickness that causes them to think that everything gets solved with tech. plants would be more profitable with better tech right? No. Same thing buffet noticed. Better computers dont make you a better investor

@mrzeal7 @koboateng @The_Ofosuhene It’s not different, it uses similar concepts. I am sure it will probably use the concept of reinforcement algorithms or classification algorithms. It Tesla can self drive, how come and Ai powered machine can’t self clean

@theandy_Ke @koboateng @The_Ofosuhene Detecting rotten tomatoes & non rotten tomatoes is vastly different from cleaning a river. AI is not self learning at the moment, because it does not have the intelligence to do so. A little sewage not acting as expected (always the case) would render that robot useless

@mrzeal7 @gwapgotme @koboateng @The_Ofosuhene I am trying to say it will work hundred percent accurate or will be precise. From the Training, you can even know its evaluation metrics. So the traditional methods won’t be abolished, they will be used together till its accuracy increases.

@theandy_Ke @gwapgotme @koboateng @The_Ofosuhene Only solution that would work is using heavily supervised 'AI' powered robot, but at that point is it still AI? But then again considering the extremely limited yet very much overblown capabilities of 'AI' currently it'd be right at home

@gwapgotme @mrzeal7 @koboateng @The_Ofosuhene Machines work faster than humans and machines can work wider area and even area with safety concerns. So building the machine may cost but it will cost more paying people to do that work on schedule times than building than machine.

@mrzeal7 @theandy_Ke @koboateng @The_Ofosuhene It would literally cost more to buy/build the tech that deals with this Than to hire a guy to clean it up. What shes suggesting costs money that no one wants to spend.

@mrzeal7 @koboateng @The_Ofosuhene Siamese are already using that technology. But some of its limitations are non interactive UI, which web app can use that. One too was limited context awareness which RAG and training of the model well can solve that.

@theandy_Ke @koboateng @The_Ofosuhene Good, how far have you gotten with it yet? If you're working on AI you should know it's limitations a lot better than most people

@mrzeal7 @koboateng @The_Ofosuhene I am trying to say, it will learn from the daily input and improves it accuracy. Haven’t you seen Ai machines that are able to detect rotten tomatoes from non rotten. The technology is already in use.

@theandy_Ke @koboateng @The_Ofosuhene AI does not train itself 'yet' please, & it probably never will. You can ask it the same question twice, it'll get it right the first time then wrong the second time. & I'm talking about the latest models. Plus operating on text is vastly different from operating on physical data

@gwapgotme @mrzeal7 @koboateng @The_Ofosuhene Naaa, the initial cost in building the machine is high but after that. It’s just maintenance cost and fuel. Assuming hiring people to clean that big ocean and using Ai powered machine, which one will cost less. And saying it’s not an issue that gets solved with tech 😂

@theandy_Ke @mrzeal7 @koboateng @The_Ofosuhene What the fuck are you saying. “Train a dataset and use mechanical means to collect waste”??it literally costs less to hire some guy to clean this app. This is not an issue that gets solved with tech lmao

@mrzeal7 @koboateng @The_Ofosuhene Yes, use can’t use Ai in a situation like this without controllers and sensors,I am currently working on a project, it uses Ai to predict fault in industries, only few companies are using this technology and they use it together with other maintenance methods.

@theandy_Ke @koboateng @The_Ofosuhene Every single 'AI powered' robot demoed as working in the real world had remote human controllers or operated in very specialized domains with very few unpredictable things capable of happening. This situation isn't one of them

@mrzeal7 @koboateng @The_Ofosuhene One thing about AI is that, it’s always not accurate and precise at its initial stage. Even with ChatGPT when it was released around 2022. It wasn’t perfect especially with image generation and its generative response. But the more it’s in use, it’s trains itself to be accurate

@theandy_Ke @koboateng @The_Ofosuhene Getting AI to work in the real world, especially in a chaotic environment like this, is something that's still more in experimental & theoretical than reliable usage stage. AI at it's current stage does not have the capacity to react to real world changes

@koboateng @The_Ofosuhene Yeah, I can work on that if only there is a dataset available for that

@theandy_Ke @The_Ofosuhene Facts

@mrzeal7 @koboateng @The_Ofosuhene You can. With supervised classification algorithms, you can train a dataset on waste in water to sort out the waste, and use mechanical means to collect the waste. That’s the work of mechanical engineers in AI 🤖

@theandy_Ke @koboateng @The_Ofosuhene You can't

You need to assist yourself to win the league.