In episode 109, we are back with Will Liang, Executive Director at MA Financial, to discuss DeepSeek and the impact on the future development of artificial intelligence and the global economy. Are tech firms going to scale back their investments due to this low cost model, or is it all a bit of a hype? Enjoy the Show!
Overview of Podcast with Will Liang on DeepSeek
01:00 First impressions of DeepSeek
03:00 Is DeepSeek really the revolution that it was made out to be, or were they not honest about the development of the model as OpenAI claims?
04:00 You need to compare apples with apples; OpenAI did not spend billions of dollars developing one model
04:30 DeepSeek used 8-bit in floating point numbers instead of 16 or 32 bit. This resulted in massive memory savings
08:00 We haven’t seen the Magnificent Seven reduce their plans for AI investments
09:00 There is a whole ecosystem around AI that might be affected
10:00 Jevons’ paradox in AI and related industries
11:00 I don’t think DeepSeek is a distilled model
13:30 Should we be worried about data being fed back to China? Apparently, there was code embedded in DeepSeek that could reroute data back to China Mobile
16:30 DeepSeek has released a lot of LLM secrets to the public
17:30 DeepSeek could really help Europe and Australia, which have been lagging behind
18:00 Applications of DeepSeek in the investment industry
22:00 You will see a lot more distilled models from DeepSeek popping up. I’ve seen five in the last week
25:00 A wake up call for the investment industry
__________
[i3] Insights is the official educational bulletin of the Investment Innovation Institute [i3]. It covers major trends and innovations in institutional investing, providing independent and thought-provoking content about pension funds, insurance companies and sovereign wealth funds across the globe. The [i3] podcast is available on Apple Podcast, Spotify, Amazon Music, YouTube Music, or your favourite podcast platform.