Amazon to invest another $4 billion in Anthropic

Pivot 5: 5 stories. 5 minutes a day. 5 days a week.

1. Amazon to invest another $4 billion in Anthropic

Amazon has announced an additional $4 billion investment in Anthropic, an artificial intelligence startup founded by ex-OpenAI research executives. The total investment is $8 billion, with Amazon remaining a minority investor.

Anthropic will use AWS Trainium and Inferentia chips to train and deploy its largest AI models. The partnership will also allow AWS customers to fine-tune their own data on Anthropic's Claude. This comes one month after Anthropic announced its Computer Use capability.

Read the full story here

2.AI can now create a replica of your personality

A team from Stanford and Google DeepMind has developed simulation agents, replicas of human subjects, to make it easier for researchers to conduct studies that would be expensive, impractical, or unethical with real human subjects. The simulation agents, which mimic human behavior, can be used to test various aspects of social media interventions and traffic jams.

The research, led by Stanford PhD student Joon Sung Park, shows that using real humans to generate personas can lead to stronger AI agents. However, the research raises questions about the ease with which people can personify others online.

Read the full story here

3. Tuvalu: The disappearing island nation recreating itself in the metaverse

Tuvalu, a Pacific Island nation, is digitally backing up its land and statehood as sea level rise threatens its shores. The government is building a virtual replica of the country, storing everything from houses to beaches to trees. The digital replica aims to preserve the nation's beauty, culture, and legal rights for generations to come.

The initiative was announced in 2022 as part of the government's Future Now project, focusing on international diplomacy and pragmatic adaptation to climate change. Tuvalu's digital twin aims to preserve its cultural heritage and protect its sovereignty.

Read the full story here

4. MIT researchers develop an efficient way to train more reliable AI agents

MIT researchers have developed an efficient algorithm for training reinforcement learning models for complex tasks with variability. The algorithm strategically selects the best tasks for training an AI agent, focusing on a smaller number of intersections that contribute the most to the algorithm's overall effectiveness.

This method maximizes performance while keeping the training cost low. The technique was found to be between five and 50 times more efficient than standard approaches on an array of simulated tasks. This approach helps the algorithm learn a better solution faster and improves performance.

Read the full story here

5. Business spending on AI surged 500% this year to $13.8 billion

Business spending on generative AI increased by 500% this year, from $2.3 billion in 2023 to $13.8 billion, according to data from Menlo Ventures. OpenAI ceded market share in enterprise AI, declining from 50% to 34%, while Anthropic doubled its market share from 12% to 24%. The power shift is due to the advancement of Claude 3.5 and the majority of companies using three or more large AI models. Meta's market share remained at 16%, while Cohere's share remained at 3%.

Google's share rose from 7% to 12%, while Mistral's lost one percentage point to 5% in 2024. Foundation models, such as OpenAI's ChatGPT, Google's Gemini, and Anthropic's Claude, still dominated enterprise spend, with large language models receiving $6.5 billion in investment.

Read the full story here