Comments by "MeTube - tacticalvote co uk" (@OneAndOnlyMe) on "Richard J Murphy" channel.

  1. 2
  2. 2
  3. 2
  4. 2
  5. 2
  6. 2
  7. 2
  8. 2
  9. 2
  10. 2
  11. 2
  12. 2
  13. 2
  14. 2
  15. 2
  16. 2
  17. 2
  18. Every business has these common functions, HR, IT, Finance, Marketing, Customer Service. In all these functions, the "operations" side can be automated because these are very standardised functions. HR Ops - 90% of the work can be automated and augmented with AI, needs only 4 to 7 people even for a business with 50,000+ employees. IT Ops - 90% of the work can be automated, needs only 10 or so people, even for a business with 1,000+ servers and a globe spanning network infrastructure. Finance Ops - 80% of the work can be automated, I recently did a reduction of a 30 person team down to 6 using automation (not even using AI, just workflow automation). Marketing Ops - 95% of the work can be automated (or outsourced entirely), needs only a small internal management layer. Customer Service Ops - 90% of the work can be automated, 9 out of 10 interactions are standardised query/response. Contact centers can be scaled up using virtual digital agents instead of hiring additional human agents. Where any job has defined processes and procedures, the workflow can be fully automated with humans only needed to handle the exceptions. This is what we are doing already today and have been for the past decade. The difference now is that the solutions are cheaper and no longer only affordable to large enterprise budgets. If you don't believe me, look at these things: ServiceNow platform (Customer Service, but also anything an organisation needs to do) Workday platform (HR and Payroll) Dynamics 365 platform (Finance and ERP) SalesForce platform (Marketing and CRM) Google AI Studio and Microsoft Co-Pilot (End user productivity tooling)
    2
  19. 2
  20. 2
  21. 2
  22. 2
  23. 2
  24. 2
  25. 2
  26. 2
  27. 2
  28. 2
  29. 2
  30. 2
  31. 2
  32. 2
  33. 2
  34. 2
  35. 2
  36. 2
  37. 2
  38. 2
  39. 2
  40. 2
  41. 2
  42. 2
  43. 2
  44. 2
  45. 2
  46. 2
  47. 2
  48. Open sourcing puts the power of LLMs in the hands of everyone. What the DeepSeek team did was more impressive than just a better LLM than OpenAI's LLM. They showed the world that you don't need ultra expensive GPUs to do the model training. America export bans Nvidia's best GPUs. But DeepSeek software engineers used a generation older Nvidia GPU, and wrote custom code in, essentially, the GPU's native machine code, to boost the performance of the older generation GPUs. The real pay off with DeepSeek is China showing the world its impressive software engineering talent. One of their engineers turned down $10 million from another Chinese tech company (Xiaomi) to work for DeepSeek. They know their worth. The US tech majors will still win from the open sourcing of LLMs. LLMs in themselves do not add value to anything, they are an academic exercise. It's when LLMs are added to software and hardware products, then value is unlocked. More so, when LLMs are successfully integrated into workflows and processes. And that is where the profits exist, in the software platforms that businesses use. Microsoft, Google, Amazon, SalesForce, ServiceNow are the big winners. (Microsoft and ServiceNow are big players in public sector, including software for government departments and administrations.) Open sourced LLMs lower the barrier to even small tech startups to get in on the game. This week's US tech stock market dip will be a blip in the mid-long term (and, yes I bought Nvidia shares on the dip at $117). It was a knee jerk reaction to DeepSeek LLM going mainstream. Already Alibaba (another Chinese tech major) has a distilled LLM that is performing even better, and is also open source. By the way, it's pronounced "en-vi-dia", not "na-vi-dia".
    2
  49. 2
  50. 2