AGI in 12 months
Andrew Tollinton
Business resilience technology to manage threats and disruption to your business
Artificial General Intelligence (AGI) in 12 months, that's the prediction of an AI consultant I work with. However, what is meant by AGI is another question - capable of doing a multitude of tasks currently performed by an employee with average IQ? Consensus estimates put AGI landing before 2028 and after AGI, comes super intelligence.
The race to develop AGI has the flavour of a nuclear arms race, with all objections sidelined by the threat of China controlled AGI. What role the UK plays is to be determined. It has opted out of the Paris AI Agreement. And, it has two of the three ingredients necessary to build large models (lacks the chips, has the PhDs and data). Having said that, it also lacks money: Trump announced a $100bn investment in Stargate every year for the next five years, its clear the UK can't go head-to-head with the US.
Many people think the launch of DeepSeek demonstrates how medium sized language models can be built economically by countries other than the US and China. 'Any force can defeat any other force, so long as it can define the battle'. We live in interesting times...
Going green gone
The last few weeks has seen climate change subjugated to the artificial intelligence race: To train artificial intelligence it's estimated US data centre energy consumption will increase from 1% of total US energy consumption to 12% by 2030.
A nuclear engineer pointed out to me there's probably not enough uranium in the world to meet these energy needs. Lets hope AGI generates more than text and images (a great carbon reduction technology would help).
As AI drives energy demand, look out for more news on another race: Nuclear fusion.
LLMs and humans will never align
AI alignment, ensuring AI produces desirable results and it doesn't tell you, you're 'a stain on the universe, please die' (Google Gemini 2024), is according to Marcus Arvan, Associate Professor of Philosophy, University of Tampa 'a buzzword, not a feasible goal'. He argues the number of possible responses LLMs will make is pretty much infinite. Therefore, the only way to control their output is to police them, just like we police humans MORE
BBCs finds LLMs are unable to accurately summarise news
The BBC asked ChatGPT, Copilot, Gemini and Perplexity to summarise 100 news stories and rated each answer. They found:
领英推荐
Measuring the impact of AI on the economy
Anthropic, the US based AI company, has launched an initiative aimed at understanding AI's effects on labor markets and the economy over time. Check out the economic indicator HERE
AI eats software businesses first
AI will take time to replace human rich roles for example, sport people, security guards etc. However, it's ability to eat software businesses has never been in doubt and now OpenAI demonstrates how online junior sales roles can be replicated by AI MORE
Security investment pays
New York City's subway has made a large investment in security and as a result, its seen 36% fewer crimes MORE
Minors used to perform hostile reconnaissance
Hostile reconnaissance by minors is underway in Ukraine. Law enforcement officials said that it detained two FSB agent groups following a?special operation in Kharkiv. These groups, per the agency, consisted exclusively of children aged 15 and 16 MORE
AI appetite
In January, I delivered two AI webinars through trade bodies involved in security. Very well attended, I noted nearly every attendee stayed until the end of the session, a clear sign that there's a real appetite for AI situation awareness.
Thank you for reading. If you found the newsletter useful please like, comment and repost.
Kind regards
Andrew
Media Solutions
1 个月Yes, we have moved on to “AGI” hype for a number of reasons, mostly due to terms of contract between OpenAI and Microsoft and all the epic fails of things marketed as being “AI” and clearly the next marketing hype term will be “ASI” with “S” standing for: supercalifragilisticexpialidocious “Even though the sound of it Is something quite atrocious If you say it loud enough You'll always sound precocious” https://m.youtube.com/watch?v=UzhE6xpQd5s
AI Implementation Strategist | Bridging Business Strategy & Technical Innovation | Graduate Teaching Assistant at Kansas State University | Helping Companies Make AI Investments That Actually Matter
1 个月Here's the AGI Blueprint https://www.dhirubhai.net/posts/christopher-royse-b624b596_agi-ai-promptengineering-activity-7294210513666748420-VVh5?utm_source=share&utm_medium=member_android