Ways to think about AGI

From Benedict Evans:

There’s an old joke that ‘AI’ is whatever doesn’t work yet, because once it works, people say ‘that’s not AI - it’s just software’.

Indeed, ‘AGI’ itself is a thought experiment, or, one could suggest, a place-holder. Hence, we have to be careful of circular definitions, and of defining something into existence, certainty or inevitably.

This fundamental uncertainty, even at the level of what we’re talking about, is perhaps why all conversations about AGI seem to turn to analogies. If you can compare this to nuclear fission then you know what to expect, and you know what to do. But this isn’t fission, or a bioweapon, or a meteorite. This is software, that might or might not turn into AGI, that might or might not have certain characteristics, some of which might be bad, and we don’t know.

By default, though, this will follow all the other waves of AI, and become ‘just’ more software and more automation. Automation has always produced frictional pain, back to the Luddites, and the UK’s Post Office scandal reminds us that you don’t need AGI for software to ruin people’s lives. LLMs will produce more pain and more scandals, but life will go on. At least, that’s the answer I prefer myself.

I’ve been wondering for a while why scrolling through LinkedIn has felt so icky lately. It wasn’t always that way for me.

Today, it finally hit me: it’s because everything looks like an ad. Even the organic posts, often from people I know, are made to look and feel like ads. It used to be that ads attempted to look like organic content, but somehow, LinkedIn managed to reverse this.

LLMs are great for replacing typing, but bad for replacing thinking.

I wonder how the business model of LLMs will evolve. Currently, they are costly to run, and most (all?) operators are losing money on them.

Some apps (like Notion) are charging extra for LLM features, which makes it much less appealing for B2C customers. Cost won’t be as much of an issue for enterprises, but that’s where data confidentiality becomes an issue.

That leaves small-ish businesses as the target market, but the scaling ambitions of LLM providers won’t be fulfilled just within that segment.

So, outside of an order-of-magnitude improvement in efficiency, what does the profitability story look like here?

Prediction of the day: in 2030, we will talk aboutt LLMs in the same way in which we talk about cloud computing today: as a foundational technology for new products and services.

Virtual Conferences Are Broken. Let’s Fix Them.

Virtual conferences should be awesome — no travel, lower prices and international by default. However, they all too often end up being just another meeting. Why is that? And how might we do better?

A picture of two dogs on a video call

What makes conferences interesting: the talks and the interactions with other participants.

In theory, virtual conferences should be great at delivering interesting talks. Speakers don’t have to travel, so it becomes easier to book the best of the best in their field. Unfortunately, that’s part of the problem. Speakers can only come up with new material every so often, so we end up with the same speakers giving the same talks over and over again.

Besides the talks, the other cool thing about conferences is that you get to meet people and can have interesting conversations with them. But this aspect also gets lost in virtual conferences. You can’t overhear conversations and jump in, you don’t have to sit at a lunch table with strangers that could become friends and you can’t spot interesting people from afar.

What remains is you sitting in a chair in your office, watching a video by yourself. It’s just another meeting.

Screw this. We must do better.

When planning the UX & Product Conference, we set ourselves the challenge to come up with a format that is more interesting than the usual sequence of keynotes. We had three requirements:

  1. It must emphasize generating new insights for everyone involved
  2. It must involve audience members as active participants (if they so desire)
  3. It must create the opportunity for serendipitous conversations

The result: a new talk format that we call 1–2-Everyone. It’s one topic, two speakers and everyone has a chance to contribute.

Here’s how it works.

  1. Each of the two speakers has 7 minutes to pitch their view on a topic.
  2. Next, the two speakers engage in a moderated debate for 15 minutes.
  3. Finally, the conversation is opened up to the audience for up to 30 minutes of discussion.

We think that this strikes a good balance. Speakers still have space to lay out their views and arguments, but the purpose of the talk-part of the format is really to set up the debate-part. Since the two speakers will challenge each other and build on their points in that debate, the topic will unfold in new and interesting ways.

Involving the audience in the discussion accomplishes multiple things. First, it makes the format more interactive and engaging. Second, it allows for a more diverse set of views to be discussed. And third, it gives visibility to individual members of the audience and to their views. That loads a trigger for conversations in the breaks or after the conference. When members of the audience express their views in this forum, it makes them easier to approach later. It’s a conversation starter.

Should 1–2-Everyone replace the classic keynotes we see in most conferences entirely? Probably not. Some topics need the space of a 45 minute talk to be laid out entirely. But we think that it will make events more interesting, more colorful and certainly less repetitive.

At the end of the day, we hope that everyone will come out of events that use 1–2-Everyone a little smarter and a little more inspired than they would otherwise.

We will debut this format at the UX & Product Conference on February 4th. We hope that other conferences will join us in adopting 1–2-Everyone, and in thinking critically about how to deliver great guest experiences in the new medium that virtual events are.

UX is a Life Skill

The world is full of design problems that just happen to be labeled as something else.

👉 Applying for a job? Treat it as a human-centered design challenge.

👉 Hosting an event? That’s just another user journey to be designed.

👉 Talking to a client? Empathize by interviewing them like a research participant.

UX is a much more versatile toolbox than we realize.
How have you used your UX skills in other areas in the past?

Why I Hire For Soft Skills First

Lists of the top 5 soft skills and the top 10 hard skills, with the annual change in ranking indicated for each skill. 4 out of 5 soft skills are at the same position, while there has been a lot of change in the hard skills

THIS is why I hire for soft skills first.

Look at this overview from the LinkedIn Workplace Learning report. Note how the top soft skills virtually don’t change year over year while the hard skills are all over the place.

It is easy to teach someone a tool or programming language. You can even learn an entirely new skill in a few weeks with the right boot camp.

But developing creativity, adaptability or collaboration skills is a lifetime’s endeavor.

So when I’m hiring, soft skills are what will make or break the deal. I’m confident you can learn the rest.

You can find the report here.

Reflection: Bias to Action

I’ve been thinking about the mantra “Bias to action” a lot this week.
It sounds great, but without context it is meaningless (like most business mantras).

What is “action”?

Is “action” to ship something?
Does running an experiment count as action?
Does conducting research count as action?

And what is it that we want to achieve with this mantra?

Is it to encourage fast learning?
Is it to avoid debating?
Is it to discourage calling out flaws in a plan and just going through with it?

These short mantras are great to create a sense of unity, but they can also create the illusion of agreement as they can be interpreted in very different ways.

It all comes down to culture: that’s the context that gives those mantras specific meaning.

Mantras can neither replace nor create culture.

Culture is the sum or your decisions.

Failure is Data

All data is equal, whether it comes from success, failure or anything in between.

If I fail but make good use of the data, I will not have failed in the grand scheme of things.

If I ignore the data in the failure, I will have failed twice.

What was the most impactful data point you found through failure?

Becoming a Generalist

When there is no guarantee that your current job will still exist a few years from now, becoming a generalist is the only rational way forward.

If you found this post on the interwebs, you probably work in a field that is evolving quickly. And if you are in such a field, you will probably benefit more from being a generalist than from being a specialist.

What do I mean by that?

As a specialist, you compete on the depth of your knowledge and skills. This is a sensible bet to make if you are in a field where the fundamentals don’t change and where deeper knowledge leads to higher rewards.

As a generalist, you compete on your ability to adapt to new contexts quickly and transfer knowledge from one field to another. This is the right bet to make if change is the norm in your field.

Thanks to my friend Thomas Gläser for triggering this thought!

Values Over Goals

Values over goals

Call me crazy, but I have mostly opted out of the goal setting game. I played that game for years and always found the quest for clear goals cumbersome at best and restricting at worst.

What do I do instead now? I focus on cultivating good values and behaviors.

I find that this focuses me more on the things within my control. Not only is that more motivating, it also gives me a useful filter for all the opportunities that present themselves along the way.

How do you operate? Has goal setting helped or hindered you in the past?

Reflection: Ability and Growth

A little reflection at the end of the week...
Growth happens when one’s ability is matched with a challenge.

I was lucky that for most of my career, the challenges that I faced stretched my abilities just enough to grow.

Early on, that happened mostly by accident and sheer dumb luck. Only later on did I start to connect the dots and seek out that channel of growth more deliberately.

But be it by luck or by design, saying yes to interesting challenges that I didn’t know how to solve yet is a pattern that has served me well for many years.

👉 What behaviors have helped you grow in your personal and professional life in the past?

Every Problem Is A Communication Problem

If we don’t live in the same reality, we can’t solve our problems.

More importantly, we won’t even know what the real problem is until we made sure that we come from the same set of assumptions and speak the same language.

We need to solve communication before we can solve anything else.

Every problem is a communication problem until we prove that it isn’t.

If People Seem Crazy, You Don't Know The Full Story

“This is crazy!”
“How could they think that!?”
“They are just stupid!”

Whenever I catch myself with thoughts like this, I know that it’s time to pause.

Few people act illogically on purpose.

If people seem crazy, you probably don't know the full story.
If people seem crazy, it’s time to get curious, shut up and listen.

Tension Is Why You Have A Job

Innovation and design are about managing tension.

The tension between refining things further and shipping early.
The tension between sticking to a vision and reacting to feedback.
The tension between the needs of the customer and the wants of a business.

Tension is good. It’s the material that we are working with.

Tension is why you have a job.

Reflection: Design Process

Reflection of the week

Reflection of the week...

I find myself worrying less and less about having clear a design process established.

I can do this because my team has really strong instincts to find the right method for the right problem at the right time. When someone doesn’t know where to go next, they ask for help. When something works, we share how and why it worked. Much of the process is emergent and ever-changing. I haven’t felt the need to draw a process diagram in more than a year. I’m really grateful for this team and it’s culture.

Because culture eats process for breakfast.

Thank you for rocking! Natalia Patiño, Nata Suu, Iñigo Hernández Muguruza, Florian (Fritz) Friesinger, Kathryn Thomas Hastings, Vinícius Krause, Lauren Nielsen, Sheeva Seyfi, Robert March, Sally Kühnlein, Kayla Svoboda, Sebastian Ploner, Jonathan Arnold

Assume Positive Intent

You can't look into people's heads. So if you have to make assumptions about their intent, you might as well assume that it's positive.

Assuming positive intent creates options, fosters collaboration, solves problems.

Assuming negative intent builds walls, feeds the ego, amplifies problems.

What do you have to lose?

What Can You Do Right Now?

What can you do right now?
I've been asking people this question in one-on-ones for years and it has always moved the conversation into a positive direction. It gets you out of discussing hypotheticals and assumptions and into discussing reality.

What are your favorite questions to move the conversation forward?

4 Learnings From Looking Back At My First Job Application

Seven years ago, I applied for a job in a slightly unconventional way: I built a website. Now that I've been a hiring manager myself for quite some time, I thought it would be fun to look back at my past self from that perspective.

Read the full article