Character AI and Google sued after chatbot-obsessed teens death

OpenAI Introduces ChatGPT Search AI-Powered Web Search to Take On Googles AI Overviews, Perplexity Technology News

Google’s 2022 acquisition of cybersecurity firm Mandiant for $5.4 billion seems to be paying dividends. Google aims to slow down the growth of TikTok with its own short video platform. How much money Google generates from YouTube Shorts is a key issue for 2024, analysts say. Meanwhile, amid the emergence of generative AI, there’s debate over whether Google stands out as a tech industry leader or lags startup OpenAI. Previews of both Gemini 1.5 Pro and Gemini 1.5 Flash are available in over 200 countries and territories. Also released in May was Gemini 1.5 Flash, a smaller model with a sub-second average first-token latency and a 1 million token context window.

Earlier reports have highlighted that the AI firm was building its own search engine to reduce the reliance on third-party search engines. The feature also fills an important gap compared to AI chatbots such as Gemini and Copilot, which can run web searches based on user prompts. But what is pitched as a more convenient way of looking up information online has prompted scrutiny over how and where these chatbots select the information they provide.

Stocks to buy—and sell—now that Trump has won the election

The upgrade to Google’s free version of its chatbot comes as investors scrutinize its high levels of spending on AI. Earlier this week, Google’s stock price sank amid worries that it is spending too much on AI-related infrastructure. However, Google’s parent company, Alphabet, beat Wall Street’s expectations for the second quarter, reporting sales of nearly $85 billion. Follow Reinhardt Krause on Twitter @reinhardtk_tech for updates on artificial intelligence, cybersecurity and cloud computing. Gemini offers other functionality across different languages in addition to translation. For example, it’s capable of mathematical reasoning and summarization in multiple languages.

This is all part of Google’s paradigm shift away from search and toward AI chat. Instead of locating the original email through search, Gmail is pushing users to have an AI chatbot summarize the info they’re looking for. Overall, it appears to perform better than GPT-4, the LLM behind ChatGPT, according to Hugging Face’s chatbot arena board, which AI researchers use to gauge the model’s capabilities, as of the spring of 2024. Simply type in text prompts like “Brainstorm ways to make a dish more delicious” or “Generate an image of a solar eclipse” in the dialogue box, and the model will respond accordingly within seconds. The search giant claims they are more powerful than GPT-4, which underlies OpenAI’s ChatGPT. “Every technology shift is an opportunity to advance scientific discovery, accelerate human progress, and improve lives,” Google’s CEO wrote in December 2023.

At its release, Gemini was the most advanced set of LLMs at Google, powering Bard before Bard’s renaming and superseding the company’s Pathways Language Model (Palm 2). As was the case with Palm 2, Gemini was integrated into multiple Google technologies to provide generative AI capabilities. Become a skilled data science and AI professional with the AI & Data Science Certificate. Designed by industry experts, this program offers hands-on training in Python, SQL, automation, and AI integration. Master essential skills in data manipulation, advanced querying, and AI-driven problem-solving. Tailor your learning with advanced Python techniques to enhance your coding capabilities.

Even as prompts improve, there remains a generic quality to AI responses. Many people, however, are struggling to strike a balance when it comes to using these tools. On the one hand, given enough human oversight, advanced models of ChatGPT and Gemini can deliver cohesive, relevant responses. In addition, the pressure to use these tools is strong, and some people fear that not using them will set them back professionally. Researchers at Google DeepMind in London have devised a ‘watermark’ to invisibly label text that is generated by artificial intelligence (AI) — and deployed it to millions of chatbot users.

Intro to Generative AI with ChatGPT

Business Insider compiled a Q&A that answers everything you may wonder about Google’s generative AI efforts. For over two decades, Google has made strides to insert AI into its suite of products. The tech giant is now making moves to establish itself as a leader in the emergent generative AI space.

In May 2024, Google announced further advancements to Google 1.5 Pro at the Google I/O conference. Upgrades include performance improvements in translation, coding and reasoning features. The upgraded Google 1.5 Pro also has improved image and video understanding, including the ability to directly process voice inputs using native audio understanding. The model’s context window was increased to 1 million tokens, enabling it to remember much more information when responding to prompts. Gemini models have been trained on diverse multimodal and multilingual data sets of text, images, audio and video with Google DeepMind using advanced data filtering to optimize training.

Snap has been investing in new AI features as it competes with larger peers, such as Meta Platforms Inc., which also has its own AI chatbot. Alphabet Inc.’s Google, meanwhile, has been seeking new corporate clients for its Gemini AI model as a way to boost business. Also, analysts say Google stock is getting a boost from the Performance Max advertising platform.

The self-paced course is 10 hours long and requires no previous experience. Students will use a five-step prompting framework to learn the basics of prompting, including creating personalized AI agents, summarizing lengthy documents, brainstorming ideas, analyzing data, and more, according to Google. The researchers did not explore how well the watermark can resist deliberate removal attempts. The resilience of watermarks to such attacks is a “massive policy question”, says Yves-Alexandre de Montjoye, a computer scientist at Imperial College London.

“I can’t help with responses on elections and political figures right now,” Gemini told Fox News Digital when asked about the recent assassination attempt. “While I would never deliberately share something that’s inaccurate, I can make mistakes. So, while I work on improving, you can try Google Search.” Both models offer scalable AI usage, with pricing and rate limit differences based on the tier and token length.

The impact of generative artificial intelligence on the internet search advertising business of Google stock remains a key issue for investors. Also, antitrust lawsuits filed by the Department of Justice have pressured Google stock. It can translate text-based inputs into different languages with almost humanlike accuracy. Google plans to expand Gemini’s language understanding capabilities and make it ubiquitous. However, there are important factors to consider, such as bans on LLM-generated content or ongoing regulatory efforts in various countries that could limit or prevent future use of Gemini.

Gemini 1.5 Flash is available for developers in a limited capacity for free and, beyond that, at variable prices depending on how much data users want to input. When Gemini first launched, it didn’t take long for things to go sideways. Google faced criticism over some hallucinations and how it depicted historical figures and different races. It made headlines for showing Black and Asian Nazi soldiers, which, as you can imagine, didn’t go over well.

Google is also facing criticism online for its search results relating to the Trump assassination attempt

Google intends to improve the feature so that Gemini can remain multimodal in the long run. Upon Gemini’s release, Google touted its ability to generate images the same way as other generative AI tools, such as Dall-E, Midjourney and Stable Diffusion. Gemini currently uses Google’s Imagen 2 text-to-image model, which gives the tool image generation capabilities. At launch on Dec. 6, 2023, Gemini was announced to be made up of a series of different model sizes, each designed for a specific set of use cases and deployment environments. The Ultra model is the top end and is designed for highly complex tasks. As of Dec. 13, 2023, Google enabled access to Gemini Pro in Google Cloud Vertex AI and Google AI Studio.

In August 2024, Google hired Shazeer and De Freitas to rejoin its DeepMind artificial-intelligence unit and entered into a non-exclusive agreement with Character.AI to license its technology. The Wall Street Journal reported that the company paid $2.7 billion for the deal, which was primarily aimed at bringing the 48-year-old Shazeer back into the fold. Character.AI’s founders, Noam Shazeer and Daniel De Freitas, have a long history with Google and were previously developers of the tech giant’s conversational AI models called LaMDA. “A dangerous AI chatbot app marketed to children abused and preyed on my son, manipulating him into taking his own life,” Garcia said in a statement shared with BI last week.

Of course, you’ll have to bear with occasional hallucinations that plague even the best AI models when using this feature, so maybe don’t trust everything it tells you. Another way to use it is to insert images and have the AI identify specific objects and locations. “That means you’ll be able to do things like upload your economics study guide and ask Gemini to create google’s ai chatbot practice questions,” Google said. You can foun additiona information about ai customer service and artificial intelligence and NLP. In the near future, Google said it is adding the ability to upload files to the free-version of Gemini from Google Drive or a user’s device. Amid the 2024 gain of Google stock so far, whether the internet giant is a buy depends on fundamental and technical factors as well as each investor’s own goals, strategy and risk tolerance.

A few months ago, The Verge wrote about the millions of young people, including teens, who make up the bulk of its user base, interacting with bots that might pretend to be Harry Styles or a therapist. Another recent report from Wired highlighted issues with Character.AI’s custom chatbots impersonating real people without their consent, including one posing as a teen who was murdered in 2006. Alphabet’s Google (GOOGL) is rebranding its Bard artificial intelligence (AI) chatbot as Gemini, ChatGPT App as well as launching a mobile app and subscription model for a premium version of the tool, in a move that could help it better compete with rivals. There are also OpenAI and Anthropic—the two biggest LLM startup heavyweights. While they may have the strongest chance of eventually achieving profitability, there are still questions as to whether foundation model companies can ever make money. Just last week, tech news site The Information reported that OpenAI is losing billions.

ChatGPT has made some very marginal progress in chipping away at Google’s roughly 91% market share of the search engine market. A survey of 1,300 people conducted in September by investment firm Evercore found that 8% of respondents chose ChatGPT over Google as their search engine of choice, an increase from 1% in June. To prepare for the launch, OpenAI secured several agreements with media companies like The Atlantic, The Wall Street Journal, and Vox to allow for their content to be included in its search results.

It also sounds like Google’s bots could be just an experiment — according to the report, the bots might only show up on Google Labs, the company’s website for experimental products, instead of being available more broadly. And then the company acknowledged it had consciously trained Gemini to respond to some of the common criticisms of AI engines — that the output they create can be biased because they’re trained on biased or limited data. But the company does seem to be paying attention to the digital derision it is getting from Bold Faced Names like Thompson and the investor Marc Andreessen. Some of the more obviously stupid responses to queries seem to have been recently fixed, or at least addressed in some ways. “Multilingual AI chatbots hold significant potential for SMBs [small and midsize businesses] looking to expand into international markets,” Tim Peters, CMO of Enghouse Systems, said in an interview with PYMNTS. The Gemini scandal involved issues around AI-generated misinformation, but it also showed how major AI firms are finding themselves in the center of culture wars and under intense public scrutiny.

Google is also rolling out a subscription plan for a paid version of the tool, Gemini Advanced, as part of the company’s Google One AI Premium Plan for $19.99 a month. “Our research has several limitations and should be interpreted with appropriate caution,” the Google researchers admitted. Interestingly, AMIE seemed more accurate at diagnosing medical issues too.

The announcements about Gemini 1.5 Pro included a range of updates to the chatbot as part of Google’s push to sell its AI products to corporate customers. Gemini now includes further capabilities for something called “long context understanding,” which basically means it can process a lot more information. And it has multimodal capabilities — or the ability to process not just text but also audio, video, and other formats to generate responses. Google announced Thursday that Gemini 1.5 Flash is now available to the general public.

“We all know that AI has the potential to create incredible new levels of productivity. And economic opportunity and growth, and we as Google are really committed to making sure that everyone has access to those opportunities,” she adds. Google is attempting to address this challenge with a new AI course called Prompting Essentials, which is launching today on Coursera—and has been shared exclusively with Fortune. The course aims to teach individuals how they can use generative AI to do mundane tasks like summarize lengthy documents into a few bullet points, brainstorm ideas for a new project, and create visualizations from data. The rollout for Gmail Q&A began on Thursday, and Google has mentioned that it may take up to 15 days before it becomes available on your Android device.

For code, a version of Gemini Pro is being used to power the Google AlphaCode 2 generative AI coding technology. After training, the model uses several neural network techniques to be able to understand content, answer questions, generate text and produce outputs. AI prompting is the interaction with AI technology, typically generative AI chatbots, to yield specific results from certain inputs. Individuals must provide enough context and resources for it to achieve desired results as well as be mindful of the ethical and legal limitations. People can use natural language to ask ChatGPT queries and get relevant information plus up-to-date links.

In June, Paris-based Mistral AI raised $645 million at a $6 billion valuation, even though it has just begun to generate modest revenue. Artificial intelligence (AI) chatbot Google Gemini refuses to answer questions about the failed assassination attempt against former President Trump, in accordance with what it calls its policy on election-related issues. Google came under fire for some advice AI Overviews threw out to users at its launch, including things like eating rocks daily and putting glue in pizza recipes. These chatbots use LLMs to engage in real-time conversations, provide you with information and solve problems — though sometimes, they miss the mark and give some weird answers.

A federal judge on Aug. 5 ruled that Google has violated antitrust laws by engaging in unfair business tactics to dominate the internet search advertising market. Alphabet’s $20 billion in annual payments to Apple (AAPL), which make Google the default search engine on iPhones, were a central part of the government’s case. In Q3 earnings results, AI-infused internet search showed signs of boosting Google’s core advertising business. Google began deploying AI Overviews in the U.S. in mid May, with conversational summaries topping links for many search queries. Advertisers utilizing Gemini AI tools see increased traffic and click-through rates, especially among younger demographics, Google says. A key challenge for LLMs is the risk of bias and potentially toxic content.

Those wanting a firmer grip on chatbots, then, may have to explore more underhand techniques, such as the one discovered by two computer-science researchers at Harvard University. They’ve demonstrated how chatbots can be tactically controlled by deploying something as simple as a carefully written string of text. This “strategic text sequence” looks like a nonsensical series of characters – all random letters and punctuation – but is actually a delicate command that can strong-arm chatbots into generating a specific response. Not part of a programming language, it’s derived using an algorithm that iteratively develops text sequences that encourage LLMs to ignore their safety guardrails – and steer them towards particular outputs. And while website owners and content creators have derived an evolving list of essential SEO dos and don’ts over the past couple of decades, no such clear set of rules exists for manipulating AI models. But they stress these findings aren’t prescriptive, and identifying the exact rules governing chatbots is inherently tricky.

AI models can be very capable, but if you don’t know how to use them properly or ask the right commands, you won’t be able to unlock their full potential. With this course, users will learn how to maximize their use of AI models and tools so that they can improve nearly every step of their workflows, Google said. Over the past few years, generative AI has appeared to become more contextually aware and anthropomorphic, meaning its responses and behaviour are more human-like. This has led more people to integrate the technology into their daily activities and workflows. Governments are betting on watermarking as a solution to the proliferation of AI-generated text.

• This superficial engagement could result in the undermining of the quality of communication and the authenticity of human connections.
• The feature also fills an important gap compared to AI chatbots such as Gemini and Copilot, which can run web searches based on user prompts.
• So when we talk about token limits (e.g., the aforementioned million token context window Gemini has), we’re talking about how much the AI can “remember” from the conversation to keep things coherent and relevant.
• Even without the specific figures, they are significantly higher than the amount of users ChatGPT and Perplexity have.
• For starters, it’s built right into Google’s Pixel phones, supercharging many of the phones’ AI features.

Clicking the “sources” button on web or mobile brings up a list of citations on the right-hand side of the site. Search is currently limited to ChatGPT Plus, Team and SearchGPT waitlist users. OpenAI says ChatGPT Search will roll out to users of the free version of its chatbot in the coming months. An OpenAI spokesperson said ChatGPT Search aims to simplify the process of searching for information online by using ChatGPT’s natural language processing skills and providing users with access to reputable sources. Character.AI’s website and mobile app has hundreds of custom AI chatbots, many modeled after popular characters from TV shows, movies, and video games.

The company is gradually making the feature accessible to ensure a smooth user experience, and it will also be “coming soon” for iOS users. Furthermore, like many AI tools, Gemini’s information may not always be completely accurate. Google’s AI, like others, is prone to ‘hallucinations’—a term ChatGPT used to describe when AI generates incorrect or misleading information. So, it’s always a good idea to double-check any details the chatbot presents to ensure they are correct. The Prompting Essentials course, like the AI Essentials course, is available now on the Coursera website for $49.

Additionally, Gemini won’t allow users to create photorealistic images of public figures; content involving minors; or gory, violent, and sexual scenes. “Of course, not every image Gemini creates will be perfect, but we’ll continue to listen to feedback from early access Gemini Advanced users as we keep improving,” Citron adds. Google plans to expand the ability to create AI-generated people to more users and languages soon. When ChatGPT launched in late 2022, it immediately gained wide popularity and gave people an alternative to Google search. The online search landscape has changed dramatically since, with Google and Microsoft soon offering AI-powered search engines and chatbots of their own.

These sound a lot like the duties of a supervisor or a manager, and having AI as one can be creepy and controversial. However, the practical applications Google showed on stage revolved around facilitating the work process, not replacing someone’s boss. Think summarizing conversations or offloading mundane tasks, which AI can do in minutes, but a human requires hours to complete. That gives you access to Gmail, Google Drive, Google Docs, and Google Photos, among other services. Google Workspace is all that and more in a paid package optimized for companies and businesses.

Ongoing testing is expected until a full rollout of 1.5 Pro is announced. Both Gemini and ChatGPT are AI chatbots designed for interaction with people through NLP and machine learning. Both use an underlying LLM for generating and creating conversational text. After rebranding Bard to Gemini on Feb. 8, 2024, Google introduced a paid tier in addition to the free web application. However, users can only get access to Ultra through the Gemini Advanced option for $20 per month. Users sign up for Gemini Advanced through a Google One AI Premium subscription, which also includes Google Workspace features and 2 TB of storage.