Amazon is investing up to $4 billion in Anthropic, an OpenAI rival, to provide advanced deep learning services to its Amazon Web Service (AWS) customers. AWS becomes Anthropic’s primary cloud provider, and this investment follows Google’s $400 million partnership with the company. Anthropic recently unveiled its consumer-facing chatbot, Claude 2, and is working on a more powerful chatbot called Claude-Next. AWS customers will be able to use Anthropic’s AI models through Amazon’s Bedrock service, positioning Amazon as a critical player in the AI field.
OpenAI’s ChatGPT chatbot now supports prompting with voice and images. Users can directly interact with the AI bot by speaking aloud or uploading a picture. The voice chat feature converts spoken questions to text and provides spoken answers, while the image search feature allows users to snap a photo and receive relevant responses. OpenAI is also introducing new models for speech-to-text and text-to-speech capabilities.
In fact, OpenAI’s GPT-4 with vision still has flaws, according to a technical paper published by OpenAI. The model struggles with making accurate inferences, hallucinates, and misses text and objects in images. It is unsuitable for spotting dangerous substances or chemicals, misidentifying certain substances, and failing in medical imaging tasks. GPT-4V also does not understand hate symbols and exhibits discriminatory behavior when production safeguards are disabled. The company acknowledges the potential risks of misuse and plans to control and restrict access to these features.
I also wanted to note the AI angle to the end of the Writers Guild of America strike. The new contract states that, per the agreement, AI cannot be used to write or rewrite scripts, and AI-generated writing cannot be considered source material, which prevents writers from losing out on writing credits due to AI. Writers can choose to use AI tools if they so desire. However, a company cannot mandate that writers use specific AI tools while working on a production. Studios must also tell writers if they are given any AI-generated materials to incorporate into a work.
Why do we care?
Listeners tagged me yesterday to comment on this (thanks, Steve!). Is a war over AI coming? Yup, and bring it on. Competition is good for those delivering solutions, as the players are invested in bringing valuable something to market. Go competition! And considering how expensive AI compute is, it appears a lot of AI is the battle of the big giants. So, you’ll understand that I will be pro-market competition here… and I’m looking at you, FTC.
Now, those feature rollouts are not without risk. Providers have an opportunity to provide services that combine AI’s capabilities with human oversight, especially in critical fields like medical imaging or security. This will include governance and compliance services related to the responsible deployment of AI,
Which is why I threw in the WGA story. Hollywood joins the AP in terms of having some basic rules for using AI. On my to-do list is adding my policies to my ethics statement. It sounds like great work to do for clients.