January 8, 2024 | Posted in News
The speed and extent to which artificial intelligence has taken hold of both the UCC space and the world at large has been incredible. Its potential is thrilling, fascinating, and terrifying in equal measure.
That isn’t to say AI emerged from a vacuum — organisations like Qualcomm and DeepMind have been researching AI for over a decade, while UCC and CX businesses like Dialpad have been working on AI-powered solutions for years. But AI only became a fully mainstream idea (practically overnight) after OpenAI’s ChatGPT burst onto the scene in 2022’s epilogue and lit the spark for its explosion into public consciousness and the catalyst for the world’s most prominent tech giants to accelerate their own AI projects.
But for most of this year, and for most managers and workers, AI tended to be thought about in abstract terms. It was a concept presented to most IT and technology admins and managers as something that would streamline workflows, boost productivity and, potentially, drive revenues while cutting costs.
OpenAI and Microsoft. Google and Anthropic. All these AI businesses put together compelling marketing pitches about how AI will revolutionise the business world and delivered some tantalising peaks into such a prospect with some major product launches. But for now, AI still remains something of an early draft of itself.
As 2023 becomes 2024, which promises to be where AI’s impact becomes palatable, where do the biggest players stand?
OpenAI started it all with the launch of ChatGPT at the tail end of 2022, introducing the average person to what AI can be in its early form.
OpenAI has come under scrutiny, however. For all its wonderful capabilities, ChatGPT is still effectively a prototype of AI at large and includes just as many flaws as any other early version of any other product. An infallible encyclopaedia of knowledge and support it most certainly is not. It’s been widely criticised for producing “hallucinations”, answers to prompts that are either inaccurate, outdated or outright lies.
Complementing the controversy of the hallucinations has been the concern about data breaches when employees input sensitive business information into their prompts, as ChatGPT, in its basic form, inputs user prompts for training. These security risks facilitated OpenAI’s decision to launch ChatGPT Enterprise in August — including enterprise-grade security and privacy, customisation options, and longer context windows for processing more extended inputs.
The launch of the faster, more advanced, subscription-based GPT-4 — whose knowledge base was up to date, rather than GPT-3.5’s limitations to being able to answer queries based on knowledge from 2021 or earlier — saw OpenAI elevate ChatGPT to another level, while ChatGPT Enterprise (and its unlimited higher-speed GPT-4 access) helped maintain its momentum over the course of 2023.
While November’s saga, in which OpenAI CEO Sam Altman was fired by its board before being reinstated four days later and all but one of the board resigning as a consequence, will have a long-term impact on OpenAI’s upward trajectory remains up in the air.
As does its closer relationship with Microsoft. Microsoft had been OpenAI’s largest investor before November’s melodrama, having put forward roughly $13 billion into the company since 2019 and having integrated ChatGPT and other OpenAI LLMs across its suite of collaboration and communications products.
Microsoft had publicly supported Altman’s return and even hired him for a few days before his return to OpenAI. Following the week of mayhem, Microsoft now has a nonvoting position on OpenAI’s board, and Microsoft’s CEO Satya Nadella has publicly spoken about monitoring OpenAI’s governance.
Although, on paper, the closer dynamic between OpenAI and Microsoft that seems to have been fostered since last month should result in greater collaboration and innovation, it’s also caught the attention of the UK’s Competition and Markets Authority (CMA), which is now probing the pair of businesses to see whether “recent developments” have seen the partnership expand into a “relevant merger situation”.
Of course, Microsoft has also been working on its own AI projects. The most significant, naturally, was Copilot — which you would probably have to have been living under a rock to have missed.
Copilot for Windows, its free, consumer-centric iteration, launched in September, while Copilot for 365, its enterprise-targeting sister, launched in November. The free generative AI productivity tool included the new Copilot user experience and Bing Chat, and if you’re reading this article on the Edge browser, you can leverage it this moment by licking the Copilot icon in the top right corner of your screen.
What 365 Copilot offers beyond Microsoft Copilot is commercial data protection, guaranteed security, privacy and compliance, the AI-powered Microsoft 365 Chat, and integration across the Microsoft 365 Apps.
Microsoft released other AI products over the course of 2023, namely Bing Chat and Bing Chat Enterprise, two generative AI tools similar to ChatGPT, albeit the latter sought to win over businesses concerned about data leaks by highlighting its emphasis on security and privacy. However, last month Microsoft announced that both Bing Chat and Bing Chat Enterprise had been rebranded under the Copilot umbrella. Copilot had become not only Microsoft’s flagship AI product but also its universal one.
Industry rumours suggest that uptake on Copilot for 365 hasn’t been as fruitful as Microsoft hoped. Potential factors could be its restriction to a minimum of 500 seats for subscription, as well as the relatively steep asking price of $30 per user per month at a time when many businesses are cutting tech budgets, given global uncertainty. However, it’s still early days, and Microsoft aims to lift the subscription seat minimum in 2024, which could boost its sales significantly.
Speaking reductively, if Google is Microsoft’s major rival among traditional tech giants in the AI race, Anthropic is OpenAI’s — to such an extent that OpenAI’s board, after firing Altman, offered the CEO gig to Anthropic’s CEO Dario Amodei, formerly an OpenAI employee. Amodei turned it down.
Anthropic is the startup behind Claude 2, a competitor chatbot to ChatGPT. Claude excels at different tasks, such as mature dialogue, elaborate content generation, complex reasoning, and detailed instruction. This is successful while maintaining a reliable and accurate end product.
Claude 2 is pitched as superior to ChatGPT by Anthropic’s backers, and the chatbot can summarise about 75,000 words, while ChatGPT can parse roughly 3,000. Users can input giant data sets into Claude 2 and request summaries in a memo, letter or story format.
Microsoft has invested billions into OpenAI, and competitors like Google, Amazon and Zoom have similarly invested serious cash in Anthropic. In October, Google announced it was investing $2 billion in Anthropic, while Zoom made an investment in May to enhance its federated approach to AI by integrating Anthropic’s AI assistant, Claude, with Zoom’s collaborative platform. Anthropic’s AI models are also now available to AWS’s customer base following Amazon’s partnership with the tech business.
While OpenAI retained most of the public’s attention, it will be fascinating to see whether Anthropic’s answer to ChatGPT fulfils its promise on a bigger stage next year. Claude 2 vs ChatGPT-4 could well be a significant rivalry in 2024.
For most of 2023, it felt like Microsoft and OpenAI were in cruise control of the AI race — not far off that famous photo of Usain Bolt smiling into the camera as he galloped to an easy win in the 100m sprint. That didn’t stop Google from making inroads, however.
Google’s generative AI ChatGPT rival, Bard, opened access in March this year.
Powered by a research large language model (LLM), Bard is a predictive engine that produces responses to prompts by selecting the words most likely to be used in conversation, as with ChatGPT. As a critical differentiator from the current, free version of ChatGPT — which has its information limited to that December 2021 deadline — Bard uses the internet to keep its information up to date and can cite the sources it extracts from.
Duet AI for Google Workspace, Google’s generative AI productivity tool for its Workspace collaboration platform, also became generally available this summer. Duet AI aims to streamline workflows by offering meeting assistance, document and conversation summaries, a chatbot for Google Chat, and personalised suggestions for Gmail responses.
While compelling products, they arguably lacked the attention-grabbing chutzpah of a ChatGPT or Copilot. Google wasn’t owning the conversation in the same way. That changed last week, however, when the giant unexpectedly launched its Gemini LLM.
Gemini, which was first announced at Google I/O in June, is now generally available to the public — specifically through an enhanced version of Bard — and is intended long-term to be integrated across virtually every Google product. Google is emphasising Gemini’s “multimodal” qualities, which means it can process and utilise different versions of data — not just text, which the average generative AI user will be most familiar with to date, but also images, code, audio and video.
Reports last month suggested that Gemini had been delayed until Q1 2024, so Gemini’s launch during its initially planned December date is something of a surprise.
Google says that Gemini Ultra is the first LLM to outperform human experts on massive multitask language understanding (MMLU). Not missing a trick, Google’s announcement blog compared Gemini’s MMLU (and other metrics) against OpenAI’s GPT-4, with its 90.0 percent MMLU beating GPT-4’s 86.4 percent.
In other words, this AI race just got very, very interesting. Gemini Ultra, Claude 2, and GPT-4. It’s going to be a fascinating 2024.
These four businesses aren’t the only major AI companies. As well as its investment in Anthropic, Zoom also released its AI Companion assistant this year, while generative AI features built into Zoom IQ were powered by OpenAI and ChatGPT. As Zoom diversified its portfolio in 2023, such as the contact centre, project management and collaborative document spaces, AI will feature in virtually everything it releases going forward.
Prominent UC players, from Cisco Webex to RingCentral to 8×8, also released exciting products and feature updates leveraging the power of AI for improved productivity and performance.
In terms of practical use cases for AI, it still feels like we’ve barely scratched the surface so far. While comparison tables and computational claims might see a company victorious in the AI battle, tangible use cases and popularity with everyday users are where the war will be won. How will Gemini, Claude 2, GPT-4, and any other unexpected LLM that might come surging into pole position meaningfully enhance people’s lives or businesses’ operational practices and financial bottom lines?