Before ChatGPT could talk, there was Alexa, but the decade-old Amazon voice assistant never reached its full potential, good only for setting kitchen timers and telling you the weather.
But now that it can harness the full power of generative AI, Amazon is turning its attention to Alexa's second act. Later this year, the company plans to release an overhauled, “more conversational” version of Alexa, sources told CNBC. In addition to the new and improved generative AI-powered features, Amazon also plans to introduce a new business model for Alexa: Alexa will require its own monthly subscription, rather than being included in the company's Prime subscription.
Alexa has so far dominated voice assistant competitors like Apple's Siri and Google Assistant, but the recent release of a wave of powerful generative AI chatbots has clearly opened up a new playing field, which Amazon says it has already sold more than 500 million Alexa-enabled devices, which could work in its favor.
At the same time, Amazon hasn't exactly presented itself as an AI disruptor: While the company is certainly supporting the AI boom through Amazon Web Services and has invested billions in and worked with AI startup Anthropic, it hasn't released any advanced AI models that would put it in the same space as OpenAI, Google, or Meta.
A generative AI-powered Alexa could be Amazon's chance to jump on the AI bandwagon and finally realize its true vision for Alexa. But the company faces real challenges: its own lagging behind in AI, the technical hurdles and rising computing costs that plague the industry, and stiff competition from other tech giants to offer the best AI assistant experience.
After recent controversy over ChatGPT's Scarlett Johansson-esque voice, OpenAI made a strong showing last week with a demo of the latest version of its product. Google has also rolled out voice support for its Gemini model, and next month Apple is expected to unveil a new, more conversational Siri featuring generative AI at its annual developers conference. Apple faces many of the same challenges as Amazon regarding its AI position, but the company is reportedly in talks with Google to bring its flagship Gemini model to the iPhone.
“Many of us had a vision for what a powerful assistant could be, but we've been held back by the underlying technology's failure to deliver on that goal,” Google CEO Sundar Pichai said. decoder “I think we have the technology that's better suited to make that happen, and that's why we're seeing progress again. It's really exciting.”
If successful, these products have the potential to completely change how we interact with technology, how we access information, how we organize our lives, and how we consume products and services. The New York Times Apple executives are reportedly concerned that new AI technology could replace the iPhone's iOS software, threatening its dominance in the smartphone market, and that an ecosystem of AI agents could eliminate the need for apps and app stores.
That's the future wearables maker Humane is hoping for with the Ai Pin, but it's obviously still a long way off. The Ai Pin was designed to talk to the wearer via voice and perform many of the same tasks as Alexa or Siri, but it hasn't lived up to expectations. Gadget critic Marques Brownlee called it “the worst product I've reviewed so far,” and Bloomberg reported yesterday that Humane is seeking an acquisition after the product's botched launch.
The dream of finally having a true AI assistant feels closer than ever, but at the same time, it still feels very far away. In OpenAI's demo last week, Sky sounded like a real human, both in voice and in answering questions (and, as always, it's important to remember that you can only trust a demo so much). But sounding like a real human is only half the battle. These models are still limited by the fact that they are always confidently wrong. As I continue to test ChatGPT, I'm amazed at how consistently wrong it is. It feels like I can't believe a word of what ChatGPT says.
“Is intelligence improving at the same rate as language ability?” asks Nilay Patel, editor-in-chief of The Verge. decoder Conversation. “To be honest, I don't think so. I think computers have become much more linguistically literate and, in some cases, stupider.”
Now for some more AI news.
Sage Lazarus
sage.lazzaro@consultant.fortune.com
sagelazzaro.com
AI in the News
News Corp signs content deal with OpenAI. The companies did not disclose terms, but of The Wall Street JournalThe News Corp-owned company reported that the deal could be worth more than $250 million over five years, including compensation in the form of cash and royalties for the use of OpenAI's technology. The partnership will allow OpenAI to use content from News Corp's consumer news publications to train models to answer users' ChatGPT queries. News Corp said BaronsMarketWatch, New York Post, The Daily Telegraphand in partnership with many publications around the world, including Axel Springer, The Associated Press, Financial TimesDotdash Meredith, and Le Monde—This transaction is perhaps the most significant to date, in view of the number of properties involved.
Microsoft and G42 will invest $1 billion in a geothermal data center in Kenya. The effort is part of Microsoft's efforts to expand its cloud services in East Africa, which also include an East Africa Innovation Lab, development and research of local-language AI models, investments in connectivity and collaboration with governments. Abu Dhabi-based G42, in which Microsoft invested $1.5 billion last month, will design and build the data centers with partners. ReutersThe companies say the plant will be “powered entirely by renewable geothermal energy” and designed with cutting-edge water-saving technology.
Nvidia reported a 262% increase in revenue, signaling continued strength around AI. The company, which makes the highly sought-after H100 GPUs needed to run today's most advanced AI models, reported quarterly sales and profits that beat expectations on Tuesday. report In particular, the data center division, which includes AI chips, achieved record quarterly sales of $22.6 billion, up 427% from the same period in 2018. The company strengthened its outlook for the current quarter, predicting second-quarter sales of $28 billion.
Colorado has reluctantly passed the state's first comprehensive AI bill. The law requires developers of “high-risk” AI systems to “use reasonable care to avoid algorithmic discrimination” and to disclose information about the systems to regulators and the public. The governor signed the bill into law on Monday, but said he explicitly asked the Legislature to replace the law with national regulation before it goes into effect in 2026. National Law ReviewAs I've reported, states have been cautious about regulating AI individually and have enthusiastically pushed for federal regulation.
Crimes using AI will lead to arrests and prosecutionsA political consultant who admitted to faked Joe Biden's voice in robocalls to New Hampshire voters has been indicted on five counts, including bribery, blackmail and oppression. NBC NewsThe FBI arrested a man this week for allegedly using the Stable Distribution Act to create “thousands” of CSAM (child sexual abuse material) images. 404 Media He then reportedly used the images to foster “an online community of like-minded criminals” and sent the material to minors. This is the first case of an FBI indictment involving the use of AI to create CSAM. MIT Technology Review Also, you published a great article. 5 ways criminals use AI.
The fate of AI
Privacy regulator questions Microsoft over upcoming Recall AI feature in Windows —David Meyer
ByteDance, Alibaba and Baidu are offering their models at ultra-low prices, intensifying the race to lead China's AI field. —Lionel Lim
PwC is using a small group of volunteer “super users” to increase employee engagement with AI. —Emma Burley
How companies can continue to add AI and other technologies without suffering from “change fatigue”—Cheryl Estrada
AI Calendar
May 29-31: GenAI Summit San Francisco 2024
June 5: FedTalks 2024 at FedScoop in Washington DC
June 25-27: 2024 IEEE Artificial Intelligence Conference (Singapore)
July 15-17Fortune Brainstorm Tech, Park City, Utah here)
July 30-31: Fortune Brainstorm AI Singapore (registered here)
August 12-14: Ai4 in Las Vegas 2024
Focus on the AI numbers
One million
Dubai plans to train this many human AI “prompters” over the next three years as part of a larger training plan, according to Khaleej Times. Dubai leaders announced the program this week, saying that while coding was once in demand, prompt engineering has become one of the most promising skills. The plan, called “One Million Prompters,” is being launched as part of Dubai's AI-advancing roadmap, the Dubai Universal Blueprint for Artificial Intelligence.
