05.23.2025

ForwardThinking 🤔

Will AI be friend or foe to workers?

What a crazy week this has been with Microsoft Build, Google I/O, Dell Technologies World, SAP Sapphire and Computex all happening at the same time. But I didn’t attend any of those lofty, news-filled gatherings. Instead I was in Cambridge, Massachusetts at the MIT Sloan CIO Symposium – a much smaller event focused on the needs of enterprise tech execs. And given the nature of this publication, it felt like exactly where I was supposed to be.

What could be better than a full day talking to executives and analysts about the changing nature of the CIO role, while watching panel after panel dive into how they’re wrestling with AI, trying to separate reality from hype, the fanciful from the practical.

The closing keynote is always a highlight at this event, and this year was no exception. Nobel Prize-winning MIT economics professor Daron Acemoglu took the stage to explore a fascinating philosophical debate in tech: will AI ultimately augment human workers, or replace them? It’s an argument that dates back to the 1950s when Alan Turing first started exploring the concept of the universal machine. Later, researchers like Norbert Wiener introduced the idea of human-machine collaboration as a counter to the argument that machines would take over everything.

Augmentation or replacement

In Acemoglu’s view, companies need to find a way to increase productivity from technology, regardless of the type, something that organizations have struggled to do. “One of the remarkable things about the age of digital technologies is that productivity growth has been very disappointing. So computers have continuously promised productivity revolutions and rarely delivered," he said.

When it comes to AI, it can go one of two ways – worker augmentation or worker replacement – and the path forward depends on a number of factors. He leaned towards the augmentation argument, but he recognized that not every manager will agree, especially when they are under constant pressure to reduce operating costs.

Photo by Possessed Photography on Unsplash

Yet he sees a way for augmentation to work for everyone if we find a way to rethink the current approaches. “It will require managers and company owners to actually understand that they can use these types of human complementary technologies, the tech sector to start becoming much more focused on these technologies and the broader public also becoming a stakeholder and thinking that perhaps a path that increases efficiency by making workers more productive, providing them more new tasks, is better for society,” he said.

What if he’s right and the promise of AI isn't in cutting costs by replacing workers with machines, but in empowering people to do more meaningful work? Workers would certainly prefer that kind of scenario to one where artificial general intelligence (AGI), however you define that, simply takes over as many tasks as possible, the cost of human capital be damned.

Workers face uncertainty

As you can imagine, the impact of AI on work has been top of mind for many workers. A Pew Research Center survey released in February, found that 52% of workers were worried about the impact of AI in the workplace. Digging further, the survey found that just 6% believe it will lead to more jobs, and that 63% report that they aren’t even using AI on the job.

This data comes at a time when CEOs are issuing an increasing number of AI-first mandates, which I wrote about in this space a couple of weeks ago. If Pew is right, most workers aren’t even experimenting with AI, yet more of their bosses are insisting that they use it on the job. That is a huge disconnect.

“One of the remarkable things about the age of digital technologies is that productivity growth has been very disappointing. So computers have continuously promised productivity revolutions and rarely delivered."

If you believe in augmentation instead of replacement, IKEA provides a prime example. The DIY store decided to use AI to enhance customer service, a story we’ve heard before. But instead of simply replacing the customer service reps, IKEA successfully retrained them as design consultants. The result? Customers get personalized help, and employees get more fulfilling, higher-value work. Truly a win-win for everyone.

Acemoglu pointed out that more companies need to look at the problem creatively as IKEA has to find solutions that increase productivity. “I think one of the very important things about the future of AI is to learn from the past of digital technology, but also dream big about how companies can revolutionize themselves in terms of offering new services, new products, new tasks,” he said. “And I don't think that's going to happen through just automating or even with dreams of AGI.”

Door with light chain in front with small Employees Only sign on it.

Photo by Erik Mclean on Unsplash

Let’s face it: generations of dystopian science fiction have conditioned us to expect the worst from artificial intelligence, but perhaps the end game isn't armies of robots taking over humanity, but instead, digital assistants taking over our email until we're out of a job. Either of those outcomes, however, is far from inevitable.

As Dropbox founder and CEO Drew Houston put it in an interview at HumanX in March, we have heard this story before with each new technological breakthrough, yet humans have always found a way to level up to the technology. "And so I think the happy version of the story is that AI is going to work for us. We get to steer it," he said.

But he also warned that you need to learn to use this technology or risk irrelevancy. “The robots are not going to take your job. It's the person who knows how to use the robot who's going to take your job,” he said. As an optimist, I choose to believe that he’s right.

~Ron

What's new on the blog 📰

Walter Sun is helping SAP build AI that understands tabular data, not just language

Walter Sun joined SAP in 2023 after almost two decades at Microsoft to head up AI at the company. It was a pivotal time, as SAP began its generative AI shift. He's helping address a number of challenges including finding a way for AI models to understand tabular data, a format in which you'll find a lot of crucial business data, especially SAP's.

Read the full article >>

SAP Head of AI Walter Sun. Image courtesy of SAP

With Neon, Databricks added a crucial database piece to the data lakehouse puzzle

When Databricks announced it was acquiring Neon last week, it filled in a crucial missing piece for the company and set it up for more dynamic agentic use cases in the future. I spoke to Sanjeev Mohan and Devin Pratt to get their perspectives on the deal.

Read the full article >>

For Salesforce’s head of security, it’s all about trust

Salesforce chief trust officer Brad Arkin not only does all of the tasks that such a role demands at a large software company like Salesforce, he also uses agents to help him and his staff work more efficiently.

That includes asking an agent for status updates instead of interrupting his busy security analysts. The agent reviews the case notes, builds timelines and summarizes progress.

“It’s like talking to somebody who’s sitting in the SOC (security operations center), who's working on the case, except me interacting with the agent isn’t distracting my staff from something they should be doing."

Read the full article >>

Photo by Ron Miller

IBM sees an AI future with smaller, more focused models in the enterprise

In April 2023, while still at TechCrunch, I had a notion that smaller language models might be better suited to the enterprise. Two years later that idea is actually beginning to gain popularity, as companies look for ways to take advantage of their own data and control ballooning AI costs. It was also a key point that IBM CEO Arvind Krishna articulated at the IBM Think keynote in Boston earlier this month.

Read the full article >>

IBM CEO Arvind Krishna at Think 2025

News of the Week 📣

New Github agent acts as developer partner, handling multiple tasks

We keep hearing a ton of agent hype, but a lot of what we've heard up until now is more aspirational than what's actually happening. But this week GitHub introduced a coding companion called Copilot coding agent. It acts as a fast, powerful assistant to handle a bunch of grunt work in the background.

"The agent starts its work when you assign a GitHub issue to Copilot or ask it to start working from Copilot Chat in VS Code. As the agent works, it pushes commits to a draft pull request, and you can track it every step of the way through the agent session logs. Developers can give feedback and ask the agent to iterate through pull request reviews," the company wrote in a press release.

Image courtesy of Microsoft

As with any AI, you would be foolish not to check its work carefully, and GitHub is careful to point out that any pull requests coming from an agent require human approval before being added to any CI/CD workflows, and that human can't be the developer who assigned the task as an extra safety measure.

That would seem essential, given these models are far from fool-proof, but this feels like a leap beyond the hype into more practical use cases, and I heard one founder of a 70-person company just gushing the other day about how powerful it was in terms of assigning multiple tasks very quickly and letting it loose.

It's worth noting that the coding agent is currently available in preview only, not generally available yet, so this is still a work in progress.

SAP announces Joule agent building studio

It was a busy week for enterprise news and part of that was because SAP was holding their annual customer conference SAP Sapphire this week in Orlando, Florida. Among the flurry of announcements coming out of Sapphire was Joule Studio, a place for customers to build custom agents on the SAP platform.

"Leveraging SAP Business Data Cloud and SAP Knowledge Graph, these agents gain semantic access to business-relevant information across both SAP and non-SAP applications, empowering them to perform their tasks with greater context and precision. As part of SAP Build, agents can also interact with workflows, automations, and custom applications, and be embedded into your own solutions to extend and enhance business processes," the company wrote in a blog post announcing the studio.

Photo courtesy of SAP

You can think of Joule like Microsoft Copilot, an all-purpose name for the suite of generative AI/agentic tools that run across the company's product set.

While nobody really agrees yet on a standard definition of agents, they are typically thought of as software that can perform a number of tasks. The agent may operate alone, or with different applications or data sets, depending on the requirements.

Joule Studio is expected to be available later this year.

Anthropic's API Update provides new agentic capabilities

You may be starting to sense a trend in this week's news, and there were a lot of agent-focused announcements this week. Anthropic, not wanting to be left out, announced two new model releases, but what I thought was the most interesting part was the agentic enhancements to the Anthropic API.

The features, which are designed to pair with Anthropic's latest model releases, Claude Opus 4 and Sonnet 4, "enable developers to build agents that execute code for advanced data analysis, connect to external systems through MCP servers, store and access files efficiently across sessions, and maintain context for up to 60 minutes with cost-effective caching—without building custom infrastructure," the company wrote in a blog post announcing the new features.

As with everything announced this week, these new features are in beta for now, as developers test them out and work out the kinks, but it does provide a way to enhance the agentic experience, at least using Anthropic's models, and that is a step forward.

Adventures in AI 🤖

Photo by Andy Kelly on Unsplash

This is a new section on how I'm using AI as I produce the FastForward blog and newsletter. Hope you like it.

This week I attended the MIT Sloan CIO Symposium in Cambridge. While I was there, I used a couple of tools to record and transcribe every interview I conducted and every panel I attended:

Otter

I’ve been using Otter for years, long before AI was cool. It falls into what I think of as 'everyday AI:' simple, useful and reliable. It handles interview transcriptions well, but where it really shines is its built-in generative AI search. I can ask questions about the conversation and get back summaries or direct quotes with timestamps.

I always double-check quotes against the transcript, but Otter is still one of my go-to tools. Another great feature is its ability to search across my entire transcript library. I can pose a question, and within a couple of minutes, it pulls relevant quotes or summaries from multiple interviews, each with a link back to the source. It’s incredibly powerful.

Granola

Granola is another transcription tool, but its strength lies in meeting summaries. It automatically highlights what was discussed and organizes it in a clean, easy-to-digest format. You can also take your own notes, and it weaves them into the summary.

It also produces a transcript, but in my experience, Otter’s is more useful for my needs. For now, I’m using both.

How are you using AI in your job? I would love to hear from you at ron@boldstart.vc.

What we're reading 📚

Person reading book, but all you can see is their hands holding the book. They are sitting on a hardwood floor steeped in shadow, creating an interesting look.

LivPhoto by Blaz Photo on Unsplash

Bloomberg

America’s Leading Alien Hunters Depend on AI to Speed Their Search

Jordan Robertson and Drake Bennett

The New Stack

Clouds, Code and Control: The New Open Source Power Struggle

Dawn Foster

FutureCIO

Build an alliance with your CIO

Jinan Budge

Medium

The Only AI Moat is Hardware, Pt. II

Murat Onen

Look who's talking 👄

"Today, 80 percent of data is in cold storage, but you can imagine a world in the not-too-distant future where data moves to warm and hot tiers, constantly in circulation, with thousands of multi-agent systems transforming millions of tasks from the data center to the edge."

- Michael Dell, this week at Dell Technologies World

Please subscribe to the newsletter and feel free to share.

3350 Virginia St, Suite 262, Miami, FL 33133

ForwardThinking 🤔

Will AI be friend or foe to workers?

Augmentation or replacement

Workers face uncertainty

What's new on the blog 📰

Walter Sun is helping SAP build AI that understands tabular data, not just language

With Neon, Databricks added a crucial database piece to the data lakehouse puzzle

For Salesforce’s head of security, it’s all about trust

IBM sees an AI future with smaller, more focused models in the enterprise

News of the Week 📣

New Github agent acts as developer partner, handling multiple tasks

SAP announces Joule agent building studio

Anthropic's API Update provides new agentic capabilities

Adventures in AI 🤖

Otter

Granola

What we're reading 📚

Bloomberg

America’s Leading Alien Hunters Depend on AI to Speed Their Search

The New Stack

Clouds, Code and Control: The New Open Source Power Struggle

FutureCIO

Build an alliance with your CIO

Medium

The Only AI Moat is Hardware, Pt. II

Look who's talking 👄

Subscribe

FastForward #23: Will AI be friend or foe to workers?

You may also like...