The AI Daily Brief: Artificial Intelligence News and Analysis - AI at Work: Is Enterprise AI the Next Big Trend?

Episode Date: June 13, 2023

Enterprise AI is a major theme for investors and corporations alike. Salesforce just announced their AI Cloud as well as doubling their generative AI venture fund to $500m. Accenture also announced a $3B investment in their AI practice. The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://theaibreakdown.beehiiv.com/subscribe Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown Join the community: bit.ly/aibreakdown Learn more: http://breakdown.network/

Today on the AI breakdown, we're talking about Salesforce's big AI announcements and the rise of Enterprise AI. Before that on the brief, a church holds an AI powered service, the EU Parliament prepares to vote on the AI Act, and much, much more. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Like, subscribe and share, and go to breakdown.network for more information. Welcome back to the AI breakdown brief. All the AI headline news you need in five minutes or less, although man is there more than five minutes of AI. AI headline news today. We kick it off with a story that has a lot of people chattering, and that's a story of a church service held in Germany that was entirely powered by AI.

The service was held as a part of the German Evangelical Church Congress, which is a biannual event held in Bavaria, Germany, and it was the brainchild of Jonas Simmerline, who is a theologian and philosopher from the University of Vienna. Simerlin worked with Chachyp.T. to come up with the themes of the sermon and then write the actual text, and then he created a set of AI avatars to actually deliver the words. Now, there was a lot of interest in this particular experiment. People started lining up hours in advance of the event, and ultimately more than 300 people were part of the service. Some gave it fairly favorable reviews, like Mark Jansen, a 31-year-old Lutheran pastor who said, I had actually imagined it to be worse, but I was positively

surprised how well it worked. But at the same time, many found the AI avatars somewhat lacking in the delivery. If you listen to my recent AI-read-Read-Sunday, you'll know that AI just doesn't always get the normal human cadence of things. And that seems to be doubly true in the context of a religious sermon. Still, it doesn't seem like the goal is a replacement of religious leaders. Simmerlene instead said that it was to show how it could help them, coming up with ideas for sermons or actually even writing them, freeing up their time for other needs within their congregations. Speaking of AI avatars, while it wasn't clear which service Simerline used for his sermon, there's a fairly good chance it was Synthesia. Now, Synthesia this morning announced

their Series C funding round of $90 million at over a $1 billion, valuation, putting them into the Unicorn Club. Their announcement blog post talks about just how far the generative AI space has come over the last five years. They write, it wasn't always this clear that generative AI would revolutionize content creation. Back in 2017, when we were still funding our AI research by selling Bitcoin, we almost went bust. We got turned down by close to 100 investors who were really only looking at predictive AI at the time. Today, they say over 50,000 businesses, which includes a third of the Fortune 100, are using Synthesia to create videos already. In addition to traditional venture firms, one of the new funders that they added in this round was

NVIDIA. Next up, we moved to a super cool story out of LA Tech Week. At that event, there was recently a virtual world hackathon, and the team that won created a fully playable role-playing game in just a single day. The reason they were able to do this was a totally different approach to programming that used AI-powered tools to move much more quickly. John Radoff writes, the core of our approach was using the Anthropic Claude LLM with its 100K context limit.

This enabled us to maintain a long story. with a high level of cohesion. The team also used Blockade Labs Skybox generator, which is something that we recently talked about on this show. One of our big themes has been the democratization of who can create what type of experience online. And while obviously these guys are programmers and already an entrepreneurial team,

the speed with which they were able to wire together AI tools to create something actually playable suggests just how different it is to create in this new environment. Speaking of creating in the new environment, Facebook has joined Google and announcing a music-focused AI genesis. generator. Invidia's Dr. Jim Fan writes meta on its impressive open source streak hits another

Lama moment for music AI. MusicGen synthesizes music audio given text or melody prompt, accompanying melatic track, or even whistling and humming. Right now, MusicGen is available as an open source model on GitHub, or you can also go to Hugging Face and use their demo interface to create a 12-second clip. As Jim mentioned, you can drop a reference audio piece or even whistle or sing into your computer microphone to give it some guidance. One of the examples they gave when they announced it was turning a Bach melody into an 80s anthem. For those of you who are interested in this sort of music creation, I'm actually thinking about a video where I compare MusicGen and Google's MusicLM pretty soon. Next up, one that we'll mention briefly because I'm going to do a full

workup later. Salesforce had their big AI announcement yesterday, and the two main features were one, they've doubled their generative AI fund from $250 million to $500 million, and two, they've announced their AI cloud, which is powered by something that they call their Einstein Trust GPD layer. Trust is, as you will see, Salesforce's bet on the big theme for enterprise AI. Now, meanwhile, over on the policy side, former British Prime Minister Tony Blair has said that AI could be the most substantial policy challenge ever. He recently co-authored a report arguing that the UK is not prepared for the challenges that lie ahead. The report called a new national purpose says, quote, AI's unpredictable development, the rate of change, and its ever

increasing power means its arrival could present the most substantial policy challenge ever faced, for which the state's existing approaches and channels are poorly configured. Now that said, the UK government did just announce that OpenAI, Deep Mind, and Anthropic had all agreed to open their models to the UK government for research and safety purposes. Remember yesterday, we discussed British Prime Minister Rishi Sunak saying that he wanted the UK to be the geographical home of global AI safety regulation. However, another contender for that title is, of course, the EU. This week after 43 technical meetings and 12 political meetings over the last year and a half,

or so, the European Parliament is slated to vote on the AI Act. The AI Act has been somewhat controversial. As Axel Voss points out, of the 85 articles, 82 were focused on the risks of AI. At an event in London last month, OpenAI CEO Sam Altman said that in its then form, they might not be able to comply with the EU AI Act. Altman went on to call the current draft of the EU Act over-regulating. Now, this set off a firestorm in the European Parliament, with many members of that body speaking out against Altman's comments. One Dutch parliamentarian said, we shouldn't let ourselves be blackmailed by American companies. If OpenAI can't comply with basic data governance, transparency, safety, and security requirements, then their systems aren't fit for the European market. Terry Breton said the rules were

not for negotiation, saying, let's be clear, our rules are put in place for the security and well-being of our citizens, and this cannot be bargained. Now, a couple days later, after all these comments, Sam Altman backtracked and said, of course, OpenAI had no plans to leave the EU. If and when the AI Act passes later this week, we will do a full workup of everything that is contained therein. For now, guys, that is it for today's AI breakdown, not so brief. If you're enjoying, please like, subscribe, and share, and I'll be back soon with the main AI breakdown. Yesterday, Salesforce held a massive AI-themed event, and today on the AI breakdown,

we're talking about not only the most important announcements from that, but what it says about the state of enterprise AI and AI at work. Welcome back to the AI breakdown. One of the big themes that you're starting to see emerge across the business sector is the rise of enterprise AI. This is, of course, AI that is retrofitted and designed for a workplace or professional use case. Now, you're seeing this in a couple different ways. First is that the leading companies that are driving the generative AI revolution are thinking about the business versions of their tools. Remember, in late April, chat GPT announced chat GPT business, which would be a subscription service both for individuals who wanted more data control, for enterprises who needed to manage multiple people, and that service was designed to be private

by default. In other words, chat GPT would not be training its future models on any conversations that flowed through that business use case. But then, of course, the big companies that have already dealt with all the challenges of becoming a B2B product vendor were inevitably going to include AI into their tools. Now, right at the very forefront of that has been Salesforce. And to be fair to Salesforce, their tools have in many ways been AI powered or at least partially AI powered for years. But at their Trailblazer DX developer conference in March, they made a number of different AI-related announcements. First, they discussed a pilot of something they called Einstein GPT, which they called, quote,

the world's first generative AI for CRM. Now, this built on an existing underlying intelligence layer they've called Einstein that has been running in Salesforce since 2016. That was more of the predictive machine learning type of variety, such as helping sales teams find the most likely next customer to buy, whereas the new generative Einstein GPT was more content-oriented, for example, helping businesses auto-generate text, pictures, and code. Clara Shee, who is then a general manager at Salesforce and who has subsequently become the CEO of AI at Salesforce, said, think of all the emails and chats that come into service agents today.

They get inundated. With Einstein GPT for service, we can auto-generate draft replies so that the agents can respond to customers much faster and they get final say. They can make any edits before they hit send. The one other big announcement from that event was the launch of a $250 million fund within Salesforce ventures that was focused on funding generative AI startups. Fast forward to yesterday, Salesforce had a major AI themed event in New York City, and there were a couple of big announcements. The first was that they were doubling the size of their generative AI fund from $250 million to $500 million.

As part of that update, they also announced investments in Humane and Tribble, which adds to other companies like Anthropic and Cohere that are already in the portfolio. Still, the bigger announcement was the launch of AI Cloud. In some ways, this is the sort of full fruition, non-based version of what they had started to announce in March. As TechCrunch put it, AI Cloud means generative AI everywhere. In fact, there are nine total GPT powered applications as part of this. There's sales GPT, which brings personalization to text generation for emails and other communications, service GPT to help with all the mundane tasks such as write-ups that surround the customer service process, marketing GPT for customizing different messages for different audience segments,

Commerce GPD for tailoring different product descriptions for particular audiences, Slack GPT, which is a no-code AI automation suite for Slack, Tableau GPT, which is a little code interpreter-esque generating visualizations from natural language prompts, Flow GPT for creating workflows from natural language prompts, and finally, Apex GPT, which is their AI coding suite. Now, more interesting than just the individual tools is, I think, the way that Salesforce is positioning their enterprise AI software. There is no doubt that the operative word is,

Trust. Their official blog post is called Salesforce announces AI Cloud, bringing trusted generative AI to the enterprise. And of course, it implies that other generative AI from other services that aren't already leaving inside your organization are not to be trusted. And boy, is this a theme reinforced throughout the post. In fact, if you search trust, there are 38 uses of trust in this particular blog post. They've even announced the Einstein GPT trust layer, which basically sits between customer data and the LLMs. It's meant to ensure that there's no potential leakage, and that enterprise customers can feel comfortable using AI technology

even for their most proprietary and sensitive needs. In his speech, Benioff said, the idea is that when our systems, when our applications, when our platforms look at all your data and then uses machine intelligence or machine learning or deep learning, we don't look at your data. We're able to provide you those predictions and that AI capability without actually looking inside the data by just keeping it anonymous.

Now, with generative AI, what we're going to be able to do is take the same technology and the same idea to create what we call a GPT trust layer. We're about to roll this out for all of our customers so they have the ability to use generative AI without sacrificing their data privacy and data security. This is critical for each and every one of our customers all over the world. Every transaction and every conversation in Salesforce begins and ends with the word trust. Now I think to some extent the place that we find ourselves and what all these announcements represent is sort of the rubber hits the road moment for AI within the enterprise, at least this generation of generative AI within the enterprise. We're now seven months deep in the post-chat GPT hype.

Individual employees have been experimenting with these different tools and bringing them into the workplace. And now you have this raft of companies, both startups as well as big established companies like Salesforce, that are trying to productize it and change workflows within the largest companies in the world. What happens next could be on the one hand, the promise transformation,

with AI being every bit as disruptive as everyone says it is, Or on the other end of the spectrum, it could end up being an overhyped flop as so often happens with new technologies. And then, of course, there's the middle possibility where a lot of it is pretty disruptive and transformational, but it's not like everything changes overnight. Along those lines, one tech analyst has said that we're getting a little bit ahead of ourselves. Dan Ives argues that AI is in its gold rush moment, but that we're closer to 1995, not 1999, as in the very, very beginnings. Still, that has not stopped money from flowing in intensely to enterprise AI-focused startups. Just in the last week or so, I've seen announcements from Contextual, which builds enterprise-focused LLMs, and just launched out of stealth with $20 million in seed funding.

Glein, which a couple weeks ago announced a nine-figure $100 million round and just introduced a workplace chat bot called Glein Chat. Business-focused co-here, which announced a $270 million series C, including a contribution from Salesforce out of that generative AI fund. And as I was preparing this video, massive consulting firm Accenture announced that it was investing $3 billion in AI to, quote, accelerate clients reinvention.

Basically, Accenture has announced that they're making a $3 billion investment over three years into their data and AI practice. That includes doubling their talent to 80,000 people, launching an AI navigator for enterprise platform, which they say can help guide AI strategy, give use cases, help with decision making and more. And they say they're going to start accelerators for data and AI readiness across 19 different industries. Paul Dardy, who's the chief

executive of the Accentra Technology Group, said, over the next decade, AI will be a megatrend, transforming industries, companies in the way we live and work as generative AI transforms 40% of all working hours. So the point of all of this is that clearly big money is being bet, that the AI hype is real and that it will transform industries in ways that are perhaps unimaginable today. Interestingly, however, this comes as there is increasing chatter that AI might be be overhyped, including for the Enterprise. A CEO of an AI company, Benu Ready, writes, Hot Take, I've pretty much stopped using Chat TPT and Bard lately and have gone back to Google search.

The results are non-reliable and or boilerplate. Even in Enterprise AI, CodeGen and Document Search are the only interesting use cases. Gen AI still has a long ways to go. So, friends, that is what we will be keeping our eyes on as more and more companies dive into this space. What do you think, is Enterprise AI real or is it just a land grab and a money grab. Let me know in the comments and if you enjoyed this video, please like, subscribe, and share it. Go subscribe to the podcast and the newsletter. Until next time, peace.