Last month, I offered a dispatch on power hungry AI that included a roundup of industry updates that covers generative AI in the workplace, mobile, cars, health and more. This month’s roundup looks at how AI is transforming industries, with rapid growth in consumer tech, shopping tools, and voice assistants.
It also covers the ethical concerns emerging, from AI-generated misinformation to the challenge of distinguishing bots from humans. I’ll continue to share the stories that we’re reading inside CR as we plot our own AI strategy — let’s dive in.
Nuts & Bolts
Billions and billions of visits to native AI engines are revealing a change in consumer behavior according to a visualization from voronoi. According to data from March 2024, ChatGPT saw 2.3B visits. This number continues to grow as Google and Bing roll out AI summaries for its users, and soon AI summaries are likely to be the norm for search results.
Google to go nuclear as it looks for new ways to run power hungry AI data centers. Google cut a deal with Kairos Power that will see the launch of the first reactor this decade with more to come by 2035. Kairos began the construction of a demonstration reactor in Tennessee. Amazon has also announced its plans to join the nuclear power game.
Apple works to catch up with the release of new iPad minis with AI writing tools, and improved Siri assistant. Apple has lagged behind other tech giants as it has tried to balance privacy concerns with consumers’ hunger for AI features. Over the next several months Apple has announced they will roll out new image generation tools, Genmoji and ChatGPT features.
Shopping
Look for ads in your AI as Google is adding them to AI overviews. As Google presents AI overviews to more consumers to prevent them from defecting to native AI engines, it is adding advertisements to these summaries to help prevent the loss of revenue it gains from standard search results. Ads will have a “Sponsored” label and Google asserts that they will help consumers connect quickly with relevant businesses.
Shop until the bots drop with Google’s Shopping AI Transformation. The shopping tool now uses AI to display the most relevant products and features dynamic filter, a personalized shopping feed to inspire you, price comparison tools, price insights, price tracking and a personalized deals section. If your shopping session gets interrupted, Google Shopping will now pick up right where you left off.
Amazon dreams of AI agents doing the shopping for you. Amazon has announced the launch of AI -generated shopping guides for hundreds of product categories and is looking to the future to autonomous AI shopping agents that can recommend items and even purchase them for you. AI shopping guides offer details on technical features, explanations of terminology and recommendations on which products to buy built on product information data, customer questions, reviews, and feedback, as well as previous buying patterns.
AI as an intelligent ad blocker. Venture Capitalist and AI expert, Jeremiah Owyang thinks AI agents could reduce the ads you see. He asserts that AI agents will perform tasks and gather desired information and “filter out extraneous information like advertisements.” This may require a shift in business models to pay for agentic services with the agents funded by a consumer subscription or paid for by a corporation that would then see transactions routed to them.
Ethics
Ethical dilemma comes to chatbots as users struggle to determine whether they should be nice to their favorite AI chatbot. As AI chatbots now respond in a normal conversational manner, some users question whether they should respond with in-kind niceties or is being nice a waste of research time. One study in Japan may have solved the question for now as it determined that impolite prompts led to increased bias and lower quality results overall.
Microsoft Copilot hallucinates and ties a list of crimes to a journalist from data that a court reporter had recorded. Journalist Martin Bernklau was shocked when he typed his name into Copilot and learned that he had escaped from a psychiatric institution, was a convicted child abuser and was preying on widows. A reminder that for the time being, answers from an AI engine should be checked and verified.
Recreated without permission, Drew Crecente received a Google Alert that his daughter’s name and picture were being used to power an AI chatbot, even though she had been murdered nearly two decades ago. He found that her name and photo was fronting a knowledgeable and friendly AI character. Crecente was appalled that this was allowed to happen and questions the AI industry’s ability to shield users from potential harm.
Turing test tipping point? The Turing Test proposed by Alan Turing in 1950 helps to determine whether a machine can behave intelligently like a human. A recent study had 500 participants engaging with 4 agents for 5 minutes each including one human agent and 3 AI chatbots. The results found that ChatGPT-4 was considered human 54% of the time, very similar to the numbers for the human agent. How will society handle this when people may no longer be able to tell if they are talking to a human or a bot?
Until next time! We hope you’ll continue to follow us here for the latest AI marketplace news roundups, and get in touch if you want to explore these updates and build solutions together.
