• CoinDesk is in the final stages of completing a $125M deal.
• Apple is developing its own AI chatbot as an internal tool.
• Societe Generale subsidiary has received France’s first crypto services license.
CoinDesk may be closing a $125M deal, according to recent reports. This news comes after Caroline Ellison’s private writings were uncovered in a legal discovery process that led to the Alameda and FTX collapse.
Apple Developing Internal Chatbot
In other news, Apple is reportedly developing its own AI chatbot as an internal tool. The chatbot will be used to help employees with their day-to-day tasks, making it easier for them to stay organized and efficient while working remotely.
Societe Generale Subsidiary Receives Crypto Services License
Societe Generale’s subsidiary has become the first company in France to receive a crypto services license from the Autorité des Marchés Financiers (AMF). With this license, it will be able to offer consulting, custody and trading services related to cryptocurrencies such as Bitcoin and Ethereum. This move by Societe Generale marks a major step forward for the adoption of cryptocurrencies in France and could open up new opportunities for investments in digital assets.
Terraform Labs Appoints New CEO
Terraform Labs has appointed Chris Amani as its new CEO, replacing Do Kwon who recently stepped down from his role at the company. Amani brings extensive experience in both public and private markets, having worked on Wall Street before joining Terraform Labs earlier this year. He will now take over from Kwon with the goal of continuing Terraform’s mission of bringing DeFi products closer to mainstream users through innovative technology solutions.
Benchmarking ChatGPT Capabilities
An op-ed published on CryptoSlate discussed how language models such as ChatGPT compare against alternatives like Anthropic’s Claude 2, Google’s Bard, and Meta’s Llama 2 when it comes to focused math reasoning capabilities. According to research conducted by StanfordUniversity & UC Berkeley, GPT-4 experienced a 97% drop in accuracy between March and June when measuring prime number accuracy due to issues following step-by-step reasoning processes; response rates dropped from 21% directly executable Python snippets also decreased due to extra non-code text appearing within responses generated by GPT-3/5/4 models during this time period too