ChatGB: Tony Blair backs push for taxpayer-funded ‘sovereign AI’ to rival ChatGPT

21 comments
  1. Controlled by the Government? Erm no thank you. Imagine asking it a question about like, ‘what Brexit a good idea?’ or ‘is Rwanda a safe country?’

  2. > applying it to NHS data, that is going to be incredibly powerful.

    Can we not? The last thing we need is AIs hallucinating diseases for us. This feels very much so like the “let’s apply Blockchain to everything” grift we saw over the last few years.

  3. > predicting it could be used by policy advisors as writing assistants and to act as virtual tutors to schoolchildren.

    I wondered what fresh hell 2023 would bring.

  4. I use chatgpt and gpt4 daily via openai and now other platforms and ms office integrations

    It’s fucking amazing

    But perhaps these idiots should ask it how to set up an ai development team and what it might cost.

    Fuck it, I’ve a tab open in my phone I’ll ask it myself.

    What does it take to set up an ai similar to chatgpt and what would be the development costs?

    Setting up an AI similar to ChatGPT involves several key steps, and the development costs can vary depending on the scale of the project and the resources available. Here’s an overview of what it takes to develop such an AI and the potential costs involved:

    Data collection and processing: You’ll need a large dataset of text to train your AI model. This could include web pages, books, articles, or any other text source. The cost of data collection and processing can range from a few thousand dollars to millions, depending on the scale and quality of the dataset.
    Model architecture and development: You’ll need to design a suitable model architecture (e.g., transformer-based models like GPT) and develop the codebase. This will require skilled AI researchers and software engineers. The cost for this can range from tens of thousands to millions of dollars, depending on the complexity of the model and the expertise required.
    Compute resources: Training a model like ChatGPT requires substantial computational resources (e.g., high-performance GPUs or TPUs). The cost of these resources can range from thousands to millions of dollars, depending on the scale of the project.
    Fine-tuning and evaluation: Once the initial model is trained, you’ll need to fine-tune it on specific tasks and evaluate its performance. This process might require additional resources like domain-specific datasets and compute power. The cost for this stage can vary from thousands to hundreds of thousands of dollars.
    Maintenance and updates: To keep the AI relevant and up-to-date, you’ll need to continually update the model with new data and improvements. This requires ongoing investment in data collection, processing, and compute resources. The cost for this can range from thousands to millions of dollars annually.
    Infrastructure and deployment: To deploy and maintain the AI in a production environment, you’ll need to invest in infrastructure, including servers, storage, and networking. The cost for this can range from tens of thousands to millions of dollars, depending on the scale and complexity of the deployment.
    In summary, the development cost of setting up an AI similar to ChatGPT can range from a few hundred thousand dollars to tens of millions of dollars, depending on the scale, quality, and complexity of the project. Keep in mind that these are approximate figures, and the actual costs could vary based on various factors like team size, expertise, geographic location, and specific project requirements.

    Me again: for additional context Microsoft sank $10 billion into OpenAi and rhat years after it was founded and had already done funding rounds and significant development

  5. Right but can we commit to more than the absolute token amounts that have been mentioned? Last report I saw was a £2m commitment. Its all well and good making these announcements but without the backing all that is ever going to happen is these sums disappearing into the pockets of consultancy agencies. I know this doesn’t gel with the Tory idea that markets will just magic up all these things for us, but we are now in an age where most of our international competitors are sinking *billions* of pounds of state money into these sectors to incubate growth, we need to be doing the same if we want to participate.

  6. We really should do it, all that will happen otherwise is at some point in the next 10 years some other country will get there first and reap all the benefits, then we’ll spend way more money than we needed to to rush our own competitor AI out.

  7. How about we sort out the glaring financial issues in society first. Before we start blowing more money up the wall on something that won’t help the general population live.

  8. A Chinese company replaced thier CEO with something similar and started making profits I’m willing to go full 40k pre-necrons can’t be any worse.

  9. In theory i do want my country to have some sort of AI but why would anyone want that under the tories? They’d just loot the economy for another hundred or so billion to give away to their friends and then if they actually managed to make one, they would code it to ensure tories always win the election or something because that is the level these traitors operate at.

  10. Country always plays catch-up.

    I was saying for years the UK could have been a pioneer in green tech. We could have been the centre of it. Then the government removed the tax breaks etc and now other countries have that market.

  11. There is no ‘tax-payer funded’ AI/computer science/tech research of groundbreaking relevance in the UK because the researcher pay is a tenth of their US salary.

Leave a Reply