Artificial Intelligence thread

[9dashline]

Your timeline only makes sense if progress were linear. The issue is that progress is geometrical. Ten years ago we were still struggling with image classification. Today, image classification is just an introductory to deep neural networks. I will give it at most five years before AI becomes capable of software engineering. 2030 is going to be the watershed moment when big-techs massively layoff their developers. Sure, we would still be programming in the 2030's, but the massive redundancies means we would be doing it at minimal wage instead of six figure salary.


Documentation doesn't generate revenues so no priority is given to it by management. Developers are overloaded with impossible deadlines that even writing good code is becoming a problem, forget about sparing time for documentation. Remember when disciplines with memory usage used to be a thing? Peppermint farm remembers. In the West, we have a pandemic of a bunch of know-nothing with arts and business degrees being in the driver seat making policies, then expect those in science and engineering to create miracles. It is a cultural problem rather than a scrum problem.

Yup, GPT 2.0 started as a next word predictor...

In less than 2 years we went from text to text , text to image, text to video, text to action and now text to game (AI nueral engine)

Chatgpt not yet 2 years old, when it first come out context window was 4k, now its 128k.... with the latest o1 showing steps ij its work and thinking etc

Now we have open source and open weight multi and omni modal models that are sota... something unimaginable just a year ago

Claude is 500k, and soon 1million tokens will be the norm

An AI upstart working on End to End AI software engineer found a way to get context window scaled up to 100 million tokens...

See AI Scientist paper

I would say in less than 2 years, Software Engineers will be out on the streets

 

[Overbom]

Qwen team is going to announce something in 19th of September. Teasing their own Strawberry-version of Qwen? Stay tuned!

Please, Log in or Register to view URLs content!

 

[9dashline]

Qwen team is going to announce something in 19th of September. Teasing their own Strawberry-version of Qwen? Stay tuned!

Please, Log in or Register to view URLs content!

OpenAI is not open at all, first its SORA lost to Kling, heck they were going to hide o1's chain of thought, and now its strawberries finna get rug pulled by China as well...with Qwen open weights the whole thing...

 

[ZeEa5KPul]

I would say in less than 2 years, Software Engineers will be out on the streets

Why don't we take a moment to review your track record of prediction before we decide how seriously to take you, hmm?

Chinese semiconductor industry

Because those GPUs like the recently restricted Nvidia A/H 100 are intended for datacenters where the buyer purchases hundreds(if not thousands) of them for massive server farms like this(with each rack holding dozens of the GPUs). I believe this is a photo of the Shenwei (Sunway)...

www.sinodefenceforum.com

 

[Overbom]

Qwen team is going to announce something in 19th of September. Teasing their own Strawberry-version of Qwen? Stay tuned!

Please, Log in or Register to view URLs content!

Seems so! Directly related to o1. ClosedAI in the mud lol

And I think it's also going to be open-weight. Huge contribution to the community

Please, Log in or Register to view URLs content!

 

[GZDRefugee]

Yup, GPT 2.0 started as a next word predictor...

In less than 2 years we went from text to text , text to image, text to video, text to action and now text to game (AI nueral engine)

Chatgpt not yet 2 years old, when it first come out context window was 4k, now its 128k.... with the latest o1 showing steps ij its work and thinking etc

Now we have open source and open weight multi and omni modal models that are sota... something unimaginable just a year ago

Claude is 500k, and soon 1million tokens will be the norm

An AI upstart working on End to End AI software engineer found a way to get context window scaled up to 100 million tokens...

See AI Scientist paper

I would say in less than 2 years, Software Engineers will be out on the streets

Mate, software engineering is way more than just programming. Code monkeys may get the boot, but how is an AI going to know when an NP-hard algorithm is infeasible given a specific system's resource constraints and an approximation is required? We do a lot of design and testing to deliver on guarantees. An AI assistant would make my life easier, taking care of grunt work so I can focus my efforts on the really difficult problems.

 

[9dashline]

Mate, software engineering is way more than just programming. Code monkeys may get the boot, but how is an AI going to know when an NP-hard algorithm is infeasible given a specific system's resource constraints and an approximation is required? We do a lot of design and testing to deliver on guarantees. An AI assistant would make my life easier, taking care of grunt work so I can focus my efforts on the really difficult problems.

Thats the same thing Go pro said before AI bots crossed the 5000 Elo threshold...

There is no part of software enginneer that won't be soon automated away within the next short years

Might as well start playing Train Simulator because thats gonna be the only engineering left when it comes to the likes of software engineering

 

[GZDRefugee]

To add a lighthearted facet to this.

 

[9dashline]

To add a lighthearted facet to this.
View attachment 135838

This will end up like the Jordan quote...

Their comeuppance will be served soon enough. And trust me, the schadenfreude from watching it unfold will be delicious.

And what he is describing is the job of a business analyst or in the age of LLM that can write prompts for other LLMs its certainly something that can be done by a BA or PM

The irony is that natural language interfacing is exactly what omnimodal LLMs are best at, so if it can replace the coding part of a software engineer, aint no reason why it cant also replace the need of having human business anaylsts, etc....

In the end its all about converting raw business logic into code... Any omnimodal AI that is allowed to truly understand a business can go straight to code, no need for middle man... even most programming languages themselves are just human machine interfacing workarounds

To that end, the entire application layer going away soon.... UI and app layer was the middle man between human interfacing with business logic, but in the age of AI its now:

Natural language -> AI -> direct to raw compute...

No need for an app layer... Just like all these apps on smartphone can be replaced by a single OS layer/level omni-modal LLM/SLM etc and no need for a separate tax app, law app, teledoc app, when the front-end is just a superintelligent LLM as API inference... no need for hundreds of different specialized individual apps...

 

[BlackWindMnt]

Mate, software engineering is way more than just programming. Code monkeys may get the boot, but how is an AI going to know when an NP-hard algorithm is infeasible given a specific system's resource constraints and an approximation is required? We do a lot of design and testing to deliver on guarantees. An AI assistant would make my life easier, taking care of grunt work so I can focus my efforts on the really difficult problems.that

An AI assistant that can be used as an replacement for google, github issues and stackoverflow that returns results within 2~3 seconds would have been usable yesterday and most likely profitable tomorrow. Lets first see LLM AI escape the benchmark games they play for VC money and create some real AI built companies where everyone in the industry think dam humans couldn't have been able to make that.

Its like graphite of the 2000s & 2010s "Graphite can do everything, except leaving the lab", i think i should borrow it and change it to "AI can do everything, except making a profit"