OpenAI’s ChatGPT has announced that as part of three big changes, it can now access current information by browsing the internet. 

Previously  

Prior to the new (Beta) change, ChatGPT had only been trained on information up until September 2021, although ChatGPT’s newer GPT-4 architecture was trained up until January 2022. This has meant that unless using a plugin, accessing current information hasn’t been possible, which has been seen by many users as one of the main weaknesses of the chatbot. 

Now 

OpenAI has announced that ChatGPT can now browse the internet to provide Plus users first (with all users to follow later) with: “current and authoritative information, complete with direct links to sources.” 

This effectively means that some ChatGPT users will soon be able to ask questions and receive up to date answers about current affairs and access current news and topics. 

How? 

OpenAI says that the ‘Browse’ feature is rolling out to all Plus users. Users may also notice that there is also a “ChatGPT September 25 Version” link at the foot of page, going to ‘Settings & Beta > Beta features’ where users can move the toggle to ‘on’ for ‘Browse with Bing’ (in the selector under GPT-4).  

The Implications 

In addition to making ChatGPT a more attractive tool to many users, this could mean that ChatGPT will take queries away from search engines and other online news sources, thereby seeing the chatbot acting as a competitor (to a degree). This will, of course, be less of a worry to Microsoft because of its close partnership with OpenAI and the fact that its Bing search will be used to enable ChatGPT to access current information. 

Two Other Changes To ChatGPT 

ChatGPT has also announced two other new capabilities for ChatGPT. As of September 25, OpenAI says it’s rolling out voice and image capabilities to Plus and Enterprise users over the next two weeks. The capabilities will enable users to ask questions and “have a voice conversation” with ChatGPT (like users of smart speakers can do e.g., Amazon Echo) or “show ChatGPT what you’re talking about” (Google’s Bard can currently do this).  

Voice 

The Voice (Beta) capability, which is being rolled out to Plus users on iOS and Android, enables users to do a number of things such as have a conversation or “speak with it on the go, request a bedtime story, or settle a dinner table debate.”  It’s interesting that in its announcement, OpenAI describes ChatGPT in this context as “your assistant,” perhaps positioning it alongside digital assistants, e.g. Amazon Alexa, Google Assistant, and Apple’s Siri. 

How? 

OpenAI says to activate it, users should head to ‘Settings → New Features’ on the mobile app and opt into voice conversations. Then, it’s a case of tapping the headphone button (top-right corner of the home screen) and choosing the preferred voice out of five different voice options. 

Images 

Open AI also says that ‘Image input’ will soon be generally available to Plus users on all platforms. This will allow users to tap the photo button to capture or choose an image, and show/upload one or more images to ChatGPT to help get answers to queries. For example, OpenAI says users can “troubleshoot why your grill won’t start, explore the contents of your fridge to plan a meal, or analyse a complex graph for work-related data.”  Image input will also enable users to focus on a specific part of the image by using a drawing tool in the mobile app. 

Challenges 

Despite ChatGPT becoming the fastest growing consumer app in history (UBS research, 2023), and OpenAI introducing these new value-adding features to the app, ChatGPT and its new features are not without their widely acknowledged challenges. For example: 

– As ChatGPT states clearly at the foot of its search page “ChatGPT may produce inaccurate information about people, places, or facts,” and its CEO, Sam Altman, has spoken freely about the chatbots ability to have ‘hallucinations’, i.e. produce content that looks plausible but is simply made up. For example, back in July, the Federal Trade Commission (FTC) sent a letter to the Microsoft-backed business requesting information on how it addresses risks to people’s reputations caused ChatGPT’s potential to “generate statements about real individuals that are false, misleading, or disparaging.” 

– As some technology commentators have noted, in addition to potentially helping to bring more creative and accessibility-focused applications, the new voice technology feature could potentially be open to misuse, e.g. malicious actors using it to impersonate public figures or commit fraud. 

– Some commentators have also noted how the new image input feature could create safety issues for users. This could include a situation when people rely on the model when it hallucinates – perhaps misreading a safety-diagram for example. That said, OpenAI has said that the model has been tested with red teamers for risk domains (e.g. extremism and scientific proficiency) and with a diverse set of alpha testers. OpenAI is also reported to have worked with the ‘Be My Eyes’ (free) mobile app for blind and low-vision people. Measures have also reportedly been taken to limit ChatGPT’s ability to make direct statements about people in its analysis of images (because it’s widely accepted that these aren’t always accurate). 

Amazon and Anthropic – Challenging Microsoft 

Just as Microsoft and OpenAI’s partnership and Microsoft’s investment have given Microsoft Copilot, and these new capabilities in ChatGPT, and Google has Bard and Duet, Amazon is now teaming up with Anthropic (which has the ‘Claude’ chatbot) to enter the generative AI world and take on Microsoft. It’s been reported that Amazon is to invest up to £3.3bn in San Francisco-based AI firm Anthropic to get Claude 2 and to create new apps and improve its existing ones for its customers. As part of the deal Anthropic will be able to leverage Amazon’s huge computing power (Amazon has the AWS cloud computing service). Chatbots typically need large amounts of computing power for their LLMs and to handle the numbers and variations of customer queries. OpenAI, for example, is able leverage Microsoft’s Azure. 

Another Perspective 

Whereas many commentators see OpenAI’s new features for ChatGPT as part of the fight-back from other tech companies against Microsoft and OpenAI (which is to be expected as companies race to offer their own value-adding version of the relatively new generative AI technology) not all agree. For example, some tech commentators have suggested that the Anthropic deal (with Amazon) is also a sign that companies like Amazon and Google are looking to challenge Nvidia’s dominance in the market for specialist AI chips. 

What Does This Mean For Your Business? 

For UK businesses navigating the rapidly evolving digital landscape, these advancements in generative AI signal an era of unparalleled access to real-time information and enhanced user engagement. OpenAI’s groundbreaking features in ChatGPT come at a time when tech giants are all recognising the commercial potential of AI-driven chatbots, a fact underscored by Amazon’s timely announcement to supercharge Alexa’s AI capabilities. Such competitive moves are not just coincidences, but they mark the onset of a race where big tech firms are vying to seamlessly integrate generative AI into their product ecosystems, a shift that will inevitably reshape how businesses and consumers interact.

In the case of ChatGPT’s competitors, these new features could have a negative effect on them, likely by taking queries away from search engines and other online news sources. 

For most UK enterprises, big tech firms vying to seamlessly integrate generative AI presents a dual-edged sword. On one hand, the ability to pull current data and have more interactive user experiences could elevate customer service, streamline operations, and drive innovation. On the other, the challenges posed by ‘hallucinations’ in AI outputs, potential misuses, and concerns over data integrity may necessitate a cautious approach. Companies, therefore, must be discerning in their adoption, weighing the transformative potential against the risks. Also, with Amazon’s massive investment in Anthropic and the resultant potential synergies with AWS, businesses may soon be faced with a broader array of AI-driven solutions, further intensifying the competitive landscape.  

As the dust begins to settle in this technological race, some would say that UK businesses stand at a crossroads, e.g. to embrace these advancements as pivotal tools for future growth, or to tread cautiously, ever mindful of the evolving implications of AI in the business realm. Others would say, on balance, using a common-sense approach and being careful to check ChatGPT’s outputs for any obvious errors, these new features and others will provide further time and cost savings cost, and efficiency, and productivity benefits to businesses as they learn the many ways they can leverage advances in generative AI and its widescale adoption.