nutanc

22.6K posts

nutanc

nutanc

@nutanc

CTO Co Founder Ozonetel

Katılım Ocak 2009
460 Takip Edilen3K Takipçiler
Sabitlenmiş Tweet
nutanc
nutanc@nutanc·
Ok, today, I will be posting multiple videos of OpenAI GPT demos I built. All of them will be MIND BLOWINGLY GOOD. Because they are demos :) But seriously, OpenAI is good for prototypes. Follow this thread if interested. At the med of the day I will do an analysis.
English
6
11
74
0
nutanc
nutanc@nutanc·
Have seen this cycle happen many times. It was with Sampras first, then Fed, then Nadal, now Djokovic. Fans hold onto their heroes creating scenarios where their man can win a few more. It doesn't matter how many more Djokovic wins or how. The game has moved on. Look at the Wimbledon final and how those people played. The speed of movement and the power of the shots. It's a new era. Djokovic matches look like something from a bygone era. He will win some because of the mental thing where others who have never beaten him can't overcome the hurdle. His problem will be the young new guys who don't care, like Fonseca showed.
Danny@DjokovicFan_

Call me crazy but I think Novak Djokovic would really benefit from having a lower ranking. At present, he can't meet Sinner and/or Alcaraz prior to the late stages of tournaments. And by then, he isn't physically fresh. His best chance to beat them and win a major is by meeting them as early as possible when he can perform at his very best. He has also proven he can beat everyone else regardless of freshness. It's just hard to beat these two when not at his best physically. Thoughts?

English
0
0
1
173
nutanc
nutanc@nutanc·
This is such a good take. It's not about AI usage is bad. It's that you didn't put the effort to make it look good. You just gave a prompt and want me to consume the slop. Nope. But the silver lining is that now I know which products care for their customers.
bob appetit@flugenhof

the problem with AI isn't just that it's ugly and lazy, it's indicative of a culture uncaring about everything. you have to care about something. you have to care about what you're putting out in the world. you have to make effort somewhere so people can perceive it.

English
0
0
1
214
nutanc
nutanc@nutanc·
Just like Claude design killed Figma and other startups, this will kill the edtech industry. Not.
Dan Fitzpatrick@theaieducatorx

BREAKING: Anthropic just launched Claude for Teachers. Complete with education-specific skills built in partnership with teachers. 🇺🇸 Teachers in the US get a year free (including Claude Code and Cowork). Claude for Teachers can connect to academic standards across all 50 states, along with curricula including Illustrative Mathematics and OpenSciEd. There is a bigger battle going on here. OpenAI made ChatGPT free for US teachers in November. Google is putting Gemini into its classroom tools. Microsoft is training educators through Elevate. At this point, free access is not really generosity, but how these companies get their tools into the daily habits of schools. Ask three questions. 1. What happens when the free year ends? By then, teachers may have created hundreds of resources and built their routines around Claude. Moving away will not feel simple, even if the eventual price is high. 2. Who decides what data can be uploaded? A FERPA compliant agreement does not answer every local question. Schools still need clear rules around student information, anonymisation and who is responsible for checking what the AI produces. 3. Do teachers actually get to keep the time they save? Research suggests regular AI users can save close to six hours a week. Great. But it means very little if leaders immediately fill those hours with extra work. I have written the full analysis for Forbes: forbes.com/sites/danfitzp…

English
0
0
2
422
nutanc
nutanc@nutanc·
@Eliana_Goldin You can continue with any gmail login(basic check to avoid bots) and you should end up at a chat interface.
English
0
0
0
10
Eliana Goldin
Eliana Goldin@Eliana_Goldin·
@nutanc That website was a bit weird when I clicked it? What is it supposed to take me to eliana@recess.gg
English
1
0
0
25
nutanc
nutanc@nutanc·
More and more that I read tweets by the recess dot gg team feels like this is the next edtech scam. I don't know why, but I just can't shake up the feeling. I might be wrong, but just feels too enthusiastic, too forced. I think almost every employee has a mandate to post something like this with just anecdotal evidence rather than actual proofs. Anyway, lets see how this one turns out.
Eliana Goldin@Eliana_Goldin

With AI, teaching will become unnecessary. But the role of teacher will become ever more irreplaceable. In the past, teachers stood at the front of the classroom and they taught. They explained things and then answered questions. Sometimes they were extra special, and they cracked jokes. Today, AI is indisputably a better teacher than human beings. It recognizes how you're thinking about a problem and addresses your particular gap immediately. It uses the Socratic Method and Scaffolded Hints without fail and never resorts to over-explaining. It speaks to you in your language, with your learning style, at your pace. And yes, it can even crack jokes. The days of frontal teaching are over. But what is not over is the need for teachers. With AI teaching the technical information, the human teacher is free to tend to the human being. This is best illustrated by example. I've worked with many kids who struggle with math. I'll give my company a quick shoutout -- we raised our students' test scores 2 standard deviations well above the mean, and in an absurdly quick amount of time. Our AI did nearly all of the math, but our tutors did something else: They broke through the emotional walls that prevent learning from happening in the first place. Our tutors noticed when math was making kids beat up on themselves. Our tutors noticed when math was no longer really about math. And that's what they spent their time on: Confidence. When @Josiah was working with his student testing in the 29th percentile, the kid would often burst into tears. Each time, Josiah would gently remind his student that he is capable, math is hard, and that if he takes it step-by-step, he can figure it out and do anything he sets his mind to. They practiced this over and over over, and one season later, the kid placed in the 65th percentile. Together, our AI Rocky optimized the curriculum and Josiah, our human tutor, healed the relationship with failure.

English
1
0
1
365
nutanc
nutanc@nutanc·
@RajivBharathan Have no clue. As I mentioned in an earlier tweet, if we go back last 20 years and see what happened to these folks, most of them and their families will still be well to do only. Dont think it affects them in any way.
English
0
0
1
5
Rajiv Bharathan
Rajiv Bharathan@RajivBharathan·
@nutanc Any idea what happens when such guys get caught ? Besides suspension etc. Does the Gov take back the wealth ?
English
1
0
0
12
nutanc
nutanc@nutanc·
It's not just this, in all probability, because @antirez has written some real good code every LLM has been trained extensively completely on his code base. So, any extensions or side projects that Antirez has tried would probably be in the distribution. I may be wrong on this, but maybe Antirez can try an out of distribution project of which he does not know anything about and then see how AI coding works?
Kai Fell@kai_fell

"Every AI-generated side project I've dabbled with worked flawlessly, so naturally that means AI works fine across every programming domain, no matter how niche, complex, or safety-critical. Reading code is a total waste of time." - Retired programmer who hasn't worked on any real software in 2 decades

English
0
0
1
249
nutanc
nutanc@nutanc·
Yep, AI is exciting. But we have to be careful not to get carried away by the excitement :) Have been experimenting with deliberate practice AI tutor based on Craig Barton principles(bot.personaas.ai/assistant/math…). Yes, human tutors are needed. But the main focus is on who do we do evals? Stuck there for now. Unless we have those, we wont have actual proofs. Would love to check out the product. Let me know if you have a sign up or something. Thanks.
English
1
0
0
38
Eliana Goldin
Eliana Goldin@Eliana_Goldin·
@nutanc I think you’re right I’m posting too much anecdotal data and not proof. Excitement is real though Happy to show you the product if you want
English
2
0
0
94
nutanc
nutanc@nutanc·
I'm a little conflicted with this. I used to think like this, but now looking at the depravity of the actions being done by the people, I don't know. By verbally attacking a dead person who has done horrific things to people might send a message to the living assholes that this is how you will be treated when you are dead. Hopefully that sends the right message to them that they should stop doing the shitty things.
Nassim Nicholas Taleb@nntaleb

The reason you do not verbally attack a dead person, no matter how vile and morally offensive he was, has to do with you & not with him. He is no threat to mankind any longer; have some skin in the game & go focus on living assholes.

English
1
0
2
203
nutanc
nutanc@nutanc·
Look at the great acting done by @SrBachchan in this episode below. Look at how immersed he feels. That kind of acting is not easy to pull off. Hats off to the man. He really is the greatest actor India produced. I mean to act in a movie like Deewar but in real life be the darpok is not easy I tell you.
Prashant Bhushan@pbhushan1

.@SrBachchan made a KBC episode on him. Amir Khan made 3 idiots inspired by him & shot partly in his school. Yet as Sonam Wangchuk has lost >8KG on his 16th day of fast at Jantar Mantar, staking his life for the future of children’s education, they ignore him!

English
0
0
0
255
nutanc
nutanc@nutanc·
Why did people use Grok CLI? Dont they know about Elon?
English
0
0
0
91
nutanc retweetledi
Joshua Khane
Joshua Khane@JoshuaKhane·
Microsoft DELETED my account AND OneDrive!!?? After ACKNOWLEDGING that I’m the owner of the account and that it was compromised??? 25 fucking years of data, thousands of euros spended on games?? My son’s baby pictures? GONE! All because MICROSOFT couldn’t bring back a compromised account?? One of the biggest companies ever coulnd’t do that so they just deleted that shit like it was nothing?? Fucking shame on you!! @microsoftnl @MicrosoftHelps @MicrosoftHelpt @Microsoft #microsoft #hacked
Joshua Khane tweet media
English
4K
8.3K
70.3K
4M
nutanc
nutanc@nutanc·
AI has brought the cost of "copying" developed software close to zero. As an example which I had shared earlier, I was able to copy the functionality of Polypad with just a prompt. Most of the features were copied. But if we ask Claude to come up with a new approach to teach kids or a new widget it cannot do that. The cost to do that is still very high. This actually opens up a can of worms. Open source licenses etc. actually don't make a lot of sense right now because we can just point to a repository and say that generate a clone of this repository with all the features. Claude has been trained well enough to say that it will not copy exactly but it will recreate the whole thing. And yes, it does recreate. But still the idea was of the owner. It's just that now it's much, much easier to generate code which works like existing code using AI. Earlier people used to fork. But then you need to adhere to original license if you fork. Should we now consider the software created by AI based on an idea in the prompt also as a fork? I dont think people will accept to that. So, in the future, we will have a lot less open source software.
Arpit Bhayani@arpit_bhayani

AI has brought the cost of development close to zero. That should make building easier. It actually makes it harder. When anything can be built, the constraint has shifted to judgment (from time or resources). Yes, clock the 120-hour weeks if that is what it takes. But that effort only counts if it moves in one direction. Motion does not mean progress, and hence hours of work in different directions make the vectors nullify :) It does not matter what level in the corporate hierarchy you operate at; learning to say no is the actual skill. At least put forth your opinion. At least it will make leadership think a bit more, or teach you a new mental model of decision-making. I feel judgment will separate the teams that build something real from the teams that just build a lot. Hope this helps.

English
2
0
5
432
nutanc
nutanc@nutanc·
First result from my teams research taking help from Claude for quick experimentation has dropped. We have been able to create a system that is 3% lesser accurate but 7 times smaller in size for an ASR system. So we have a usable ASR system that is less than 50MB in size. More importantly we have found an approach to distill bigger models. Will clean up and write a paper on it. Will share the approach also tomorrow once the remaining experiments land.
English
0
1
9
1.3K