ChatGPT Gets a Voice; Amazon Acquires Its Rival
The acquisition cost $4 billion at a time when the OpenAI gave its baby the power to talk shop
The battle for owning generative AI (GenAI) or at least put more use cases up there has gone to the next level with Amazon investing $4 billion to acquire Anthropic on the same day that OpenAI stole a march over its fledgling rival by adding voice and image smarts to ChatGPT.
This growing affinity amongst the Big Tech to acquire smart technology startups could worry the doomsday predictors, who had voiced their concerns on how those with big purses may end up swallowing up others with bigger ambitions to make money. Sadly, that’s how the global business has functioned in the tech space for decades now. The Big Fish get the Small Ones.
Amazon is seeking to grow its AI war-chest
For Amazon, the move represents a “me-too” approach as arch rivals Google, Microsoft, NVidia and Meta have already shown their Gen AI cards. Anthropic, a start-up that operates an AI-powered text analyzing chatbot (not another one!). The company is pumping in $1.15 billion for a minority stake with the option to hike it to $4 billion.
If Microsoft has OpenAI and Google has the Bard, Amazon too wants one. Hence Anthropic. It remains to be seen how Google reacts to the latest move as the company pumped in early investments into Anthropic. The company launched chatbot Claude 2 earlier this month while revealing plans to build Claude-Next that it said would be more powerful than most AI.
So, where did Amazon actually fit in? For starters, Anthropic has been an AWS customer since 2021 and now Amazon possibly sees a synergy that could help it enhance customer experiences. What’s interesting is that Anthropic has long-standing relationships with Salesforce and Zoom among others.
Andy Jassy’s statement post the announcement said, “Customers are quite excited about Amazon Bedrock, AWS’s new managed service that enables companies to use various foundation models to build generative AI applications on top of, as well as AWS Trainium, AWS’s AI training chip, and our collaboration with Anthropic should help customers get even more value from these two capabilities.”
What’s in it for Anthropic though?
From Anthropic’s point of view, Amazon’s backing allows it the muscle to raise more funds, which is a prerequisite for genAI research and use-case development. OpenAI could push the envelope only after Microsoft invested a whopping $11 billion over the years. The company has committed to provide AWS customers with access to its future models via Amazon Bedrock, the fully managed service providing secure access to top foundation models.
Training state-of-the-art models requires extensive resources including compute power and research programs. Amazon’s investment and supply of AWS Trainium and Inferentia technology will ensure we’re equipped to continue advancing the frontier of AI safety and research,” Anthropic’s statement said.
OpenAI steals a march with voice chats
And while all this was playing out, OpenAI quietly (well, not quite so!) added features that would make its popular GenAI assistant more interactive through voice conversations. Now a user can verbally ask ChatGPT to create a bedtime story on the fly with some prompts to guide the narrative. Or it can ask questions to which the chatbot can respond with voice.
That’s not all, as users would also be able to search answers using images. Imagine how cool it would be if you can upload a picture and ask the chatbot for explanations. Well, the latest update has allowed ChatGPT to do so. OpenAI revealed that it tied up with voice actors to create five different options.
What’s more, the voice responses are then parsed through its open source Whisper speech recognition system to transcribe it into text. The company also revealed that Spotify would be a launch partner that will introduce a new feature for podcasters who can sample their voice and then translate their shows from English to Spanish, French and German – all in their own voice!
An example of self-regulation by OpenAI
One factor that stood out in the announcement is that OpenAI isn’t making the technology available to anyone other than the five voice artists. In a blog post, OpenAI says the new voice tech can craft realistic synthetic voices from a few seconds of real speech. However, these capabilities also present new risks, such as potential for malicious actors to impersonate public figures and commit fraud, the post notes.
These features would roll out to paying Plus and Enterprise subscribers over the next two weeks and users can activate them by going to the “settings” and “new features” where they can opt into voice conversations. They need to click on the headphone button and select the voice they want. For now it is being limited to ChatGPT iOS and Android while image search will get enabled on all platforms by default.
