All things AI.

BlueMoonAcrossThePond said:
IMO C/C++ are doomed in the long run - because the C and C++ design committees are either intrinsically incapable of offering a safe language (C) and/or prioritize speed over safety and progress towards safety is handicapped by backwards compatibility and lack of focus (C++).
Click to expand...

stare-angry.gif

The Rust revolution and its consequences have been a disaster for the human race Linux kernel.
 
Damocles said:
It was a reference to some conversations I was having on here about 8 years ago and was called a mad doomsayer

forums.bluemoon-mcfc.co.uk

Robotics/AI

I think it's bigger than all of those. I think Robotics will be a social revolution of the size of only two others in terms of change - the Agricultural Revolution which essentially invented the concept of cities and the Industrial Revolution which invented the concept of mechanical labour. I...
forums.bluemoon-mcfc.co.uk forums.bluemoon-mcfc.co.uk

Which in itself is a reference to a 10 year old discussion here:

forums.bluemoon-mcfc.co.uk

Can people pick holes in this please?

I see the fact that we're broadly rewarded for the benefit we confer upon others to be quite egalitarian in a sense. Why shouldn't a footballer, who's had little or no formal education, be paid insane sums of money if he brings a degree of happiness to millions? Why shouldn't a plumber earn...
forums.bluemoon-mcfc.co.uk forums.bluemoon-mcfc.co.uk

I've seen this coming for a decade and started transitioning careers because of it. The "It's going to kill us all with absolutely no escape" thing was something I realised about 6 or 7 years back
Click to expand...

What did you transition from and to Damo curious?
 
Damocles said:
View attachment 156400

The Rust revolution and its consequences have been a disaster for the human race Linux kernel.
Click to expand...
And yet Torvalds backs Rust.

There's plenty of talented, but extremely opinionated and unreasonable personalities, in the Linux development sphere. Torvalds has had to become personally involved in the bullshit back-and-forth arguments regarding Rust in Linux - and has landed squarely on the side of Rust.
 
grunge said:
On a different note, there really does seem to be a situateion where the C level teams at companies believe that AI is currently more advanced that it is and are starting to rely on it far to much, a client of ours at the CEO level honestly believes that he can make any software he needs himself just using Cursor. Thankfully he tried it and failed horrifically.

For me there are 3 groups of companies. those trying to sprint before they can walk, they will over rely on it and will fail because of it, and we just have to hope they dont cause significant damage when failing, prime example would be a company that makes dangerous stuff like Nuclear reactors cutting corners on engineeers and trusting what AI gives them and it causes a meltdown.

there will be companies that learn to walk before they run and make a success of it.

and then there will be the cautious companies. those who avoid AI due to risks of data leaks etc and just get left so far behind the curve they cant catch up. again another client of ours has a risk of this, they had a hard NO policy on AI for the last 18 months. its only in the last month or 2 they have started to allow it to be used.
Click to expand...

I generally don’t disagree with your assessment there are many people who fundamentally misunderstand the technology and its current limitations. I don’t doubt that eventually AI will get to a stage where coding becomes a bit of a redundant skill but it’s certainly not there yet.

That said, it has been a massive accelerator for me personally - and I imagine also for others who know just enough to be dangerous. I have coding experience in C++, python, a bit of javascript but I’m certainly not a developer. I work in more of a data product capacity, and sometimes what I want is to build a simple proof of concept that I can show to business stakeholders, gather their feedback, and then pass that POC onto the more technical folks so they know exactly what we’re aiming for and they can then make sure it is built out with sufficient rigour. But it gives them a much more tangible starting point than just taking a list of requirements and trying to cook something up which business people see and then go “it’s not really how I imagined it” and then they have to go and do it again.

If I wanted to do this before I would have needed a software engineer and maybe a data engineer to build a POC for me. I don’t need that any more, it’s one less thing for them to do and one less obstacle for building a decent product.

I’ve also built out automated dashboards and data visuals for things like my home budgeting that previously would have been manual or botched together in Excel. The quality of process and output I’m able to achieve with this stuff is an order of magnitude better with an AI supporting.

As with all tools it’s about knowing how to use it but more importantly, how not to use it.

Like Damocles, I also have realised some time ago that I’m on a road to being completely unnecessary to businesses and so I’m in the process of pursuing a career change, though admittedly I’m not sure there is a “safe” job at the moment. There are only jobs which might give you a bit more time if you’re lucky.
 
I have transitioned from doing to overseeing the machines. Eventually the machines won't need me.
Everyone has to let the machines do and become an overseer. If you continue to do you will find yourself out of work.
We all have to adapt like mankind has done before.
 
kippax_blueboy said:
AI artificial intelligence or just intelligence. I think the tech bro's will keep developing this tech until it can do the work of everyone. Maybe 10% - 20% of all work now but in 5 years time who knows. Are we ready as a society for self driving trucks and buses. Robot surgeons, fully automated call centres, robots in the home as cleaners.
Click to expand...
As someone said, your job won't be taken by AI, it'll be taken by a person who knows how to use AI better than you do.
 
SkyBlueFlux said:
I generally don’t disagree with your assessment there are many people who fundamentally misunderstand the technology and its current limitations. I don’t doubt that eventually AI will get to a stage where coding becomes a bit of a redundant skill but it’s certainly not there yet.

That said, it has been a massive accelerator for me personally - and I imagine also for others who know just enough to be dangerous. I have coding experience in C++, python, a bit of javascript but I’m certainly not a developer. I work in more of a data product capacity, and sometimes what I want is to build a simple proof of concept that I can show to business stakeholders, gather their feedback, and then pass that POC onto the more technical folks so they know exactly what we’re aiming for and they can then make sure it is built out with sufficient rigour. But it gives them a much more tangible starting point than just taking a list of requirements and trying to cook something up which business people see and then go “it’s not really how I imagined it” and then they have to go and do it again.

If I wanted to do this before I would have needed a software engineer and maybe a data engineer to build a POC for me. I don’t need that any more, it’s one less thing for them to do and one less obstacle for building a decent product.

I’ve also built out automated dashboards and data visuals for things like my home budgeting that previously would have been manual or botched together in Excel. The quality of process and output I’m able to achieve with this stuff is an order of magnitude better with an AI supporting.

As with all tools it’s about knowing how to use it but more importantly, how not to use it.

Like Damocles, I also have realised some time ago that I’m on a road to being completely unnecessary to businesses and so I’m in the process of pursuing a career change, though admittedly I’m not sure there is a “safe” job at the moment. There are only jobs which might give you a bit more time if you’re lucky.
Click to expand...

I think one of the best descriptions for this at the current time is that you can hit MVP very quickly but the tech debt ramps up equally fast.

I knocked up a quick chatbot style interface in Kotlin+Compose and Swift+SwiftUI for Android/IOS in a few hours last night that normally would have take a few days but not close to ready to go but proved the concept of what I was aiming for

will still take time to get it production level but as a proof of concept I cant fault that one.
 
BlueMoonAcrossThePond said:
And yet Torvalds backs Rust.

There's plenty of talented, but extremely opinionated and unreasonable personalities, in the Linux development sphere. Torvalds has had to become personally involved in the bullshit back-and-forth arguments regarding Rust in Linux - and has landed squarely on the side of Rust.
Click to expand...

Linus doesn't back Rust, he's just not stopping it. In fact he's complained about it numerous times.

Attempting to replace C with Rust in the kernel is a spectacularly stupid idea.
 
Has to be said the new Google veo3 AI video is good. Still a bit of uncanny valley to it but it’s very close
 
grunge said:
Google IO yesterday was all AI by the look of it. used to be 90% Android.

Click to expand...


Text summary below... And yes it does sound like it was almost purely AI. The instantaneous translation is nice and the synth ID detector is interesting. Not that exciting otherwise. I'm worried about the internet being filled with AI videos though. The amount of crappy AI generated images is so annoying.



This Google I/O 2025 keynote covered a variety of advancements and new features across different Google products.

Here are the main talking points from the video:

  • Gemini Updates: Several updates to Gemini were announced, including:
    • An updated version of Gemini 2.5 Pro, described as their most intelligent model.
    • An updated Gemini 2.5 Flash with improvements in reasoning, code, and long context, with general availability expected in early June.
    • Introduction of Deep Think mode for Gemini Pro to push performance limits [04:30].
    • Extension of Gemini to be a "world model" capable of planning and imagining [04:55].
  • Google Beam: A new AI-first video communications platform that transforms 2D video into a realistic 3D experience using multiple cameras and AI rendering on a 3D light field display [00:23, 00:56]. Early devices from a collaboration with HP are expected later this year [00:56, 01:04].
  • Google Meet: Real-time speech translation is now available in English and Spanish for subscribers, with more languages and enterprise availability coming later this year [01:04, 01:22].
  • Project Mariner: A research prototype agent that can interact with the web to perform tasks, demonstrated by finding an apartment [01:22, 01:44]. An experimental version will be available soon for Gemini app subscribers [02:00].
  • Personal Context: Gemini models can use relevant context from your Google apps (like Gmail) with your permission to enable features like personalized smart replies [02:07, 02:30].
  • Coding Enhancements:
    • Gemini 2.5 Pro's coding capabilities were highlighted, including the generation of a simple web app [03:50, 04:02].
    • Jules, an asynchronous coding agent for complex tasks, is now in public beta [04:12, 04:24].
  • Text to Speech: New previews with multi-speaker support for two voices, working in over 24 languages with easy language switching, are available in the Gemini API [03:14, 03:41].
  • AI in Search:
    • A new AI Mode is coming to Google Search in the US, offering comprehensive responses with links and personalized suggestions [05:31, 06:12].
    • Deep Search uses advanced query techniques to generate expert-level, cited reports [06:27, 06:33].
    • Search Live integrates Project Astra's live capabilities, allowing search using your camera for real-time information [06:51, 06:59].
    • A new try-on feature for clothing in search uses a custom image generation model to show how styles might look on you [07:13, 08:08].
  • Gemini App Updates:
    • Gemini Live now includes camera and screen sharing, and works in over 45 languages and 150 countries, rolling out for free on Android and iOS [08:36, 08:57].
    • Deep Research will soon allow uploading personal files and researching across Google Drive and Gmail [08:57, 09:04].
    • Canvas, Gemini's interactive co-creation space, can transform reports into web pages, infographics, quizzes, and podcasts [09:12, 09:21].
    • Gemini in Chrome will be an AI assistant integrated into the browser, understanding the context of the page [09:29, 09:44].
    • The latest image generation model, Imagine 4, is coming to the Gemini app [09:52, 10:07].
  • New Models for Media Creation:
    • V3 is a new state-of-the-art model with native audio generation [10:22, 10:34].
    • Lyria 2 can generate high-fidelity music and professional-grade audio [10:58, 11:07].
    • A new Synth ID detector can identify invisible watermarks in generated media (image, audio, text, video) [11:17, 11:33].
    • A new AI filmmaking tool called Flow, combining VO, Imagine, and Gemini, is launching [11:39, 12:03].
  • Android XR: Google is building Android XR for various XR devices in collaboration with Samsung and Qualcomm [12:43, 13:00]. Samsung's Project Muhan, the first Android XR device, will be available later this year [13:00, 13:12]. The glasses are designed for all-day wear with features like cameras, microphones, speakers, and an optional in-lens display, working with your phone [13:12, 13:47]. Eyewear partners for Android XR include Gentle Monster and Warby Parker [15:22].
Particularly Interesting Points:

  • Google Beam's 3D video communication [00:35] presents a novel approach to video calls that could revolutionize remote interactions.
  • The advancements in AI-powered search, specifically the live camera integration and the virtual try-on feature for clothes [06:51, 07:32], demonstrate significant improvements in the user experience by making search more interactive and personalized.
  • The development of Android XR and the demonstration of its capabilities through smart glasses [12:43, 13:47] offer a glimpse into a future where AI assistants are seamlessly integrated into our daily lives and visual interactions with the world.
  • The introduction of models like V3 with native audio generation and Lyria 2 for high-fidelity music [10:22, 10:58] highlights substantial progress in AI-generated media, expanding creative possibilities.
  • The Synth ID detector [11:25] stands out as a crucial tool for addressing the growing challenge of identifying AI-generated content, enhancing transparency and trust in digital media.
 
I quite liked the XR stuff that you can ask what you saw earlier in the day. “What was the name of that coffee shop I got coffee from”

Bit black mirror tho
 
Veo3, Seems as though results vary, Some very good, hard to tell its AI. Others a still quite poor around the mouth when speaking.

 
On a different note, for the love of god dont trust Google AI overview in there search engine.


Random test. Chris Pratt has never been on Big Bang....

1748007968911.png

2 mins later.
1748007991707.png
 
It often makes me wonder whether to keep a room in the house that is technology free (without Alexas or Phones in, or anything with a mic to be honest). As I am sure everybody else does, get bombarded with adverts (within seconds) associated with whatever you and your other half have discussed,
 
Need to add to the craziness of some of the programming too. My youtube account is the main one on my computer, my phone, the tv, my Mrs's computer. She spends an enormous amount of time on my youtube accounts, but yet it does not seems to register or add to AI recommended lists anything she searches for. (Of course it is in history). but nothing ever comes on recommended for the stuff she watches.
 
SamTheGuru said:
Need to add to the craziness of some of the programming too. My youtube account is the main one on my computer, my phone, the tv, my Mrs's computer. She spends an enormous amount of time on my youtube accounts, but yet it does not seems to register or add to AI recommended lists anything she searches for. (Of course it is in history). but nothing ever comes on recommended for the stuff she watches.
Click to expand...

I don't think it's as clever as we think. We get bombarded with more adverts than you realise and some just happen to be relevant.

Like I actually got an advert for a product I'd searched for which is unusual but it makes sense.

What doesn't make sense is the constant adverts I get for follow on milk despite there not being a baby in the house and nothing coming soon etc.
 
You must log in or register to reply here.
Unread Posts

Don't have an account? Register now and see fewer ads!

SIGN UP
Back
Top