AI Assistance - What remains after the dust has settled

AI Assistance - What remains after the dust has settled

This essay has been written 100% AI free.

Disclaimer: The author of this article is trying to predict the future. Although I spent some time to think this through, this is very thin ice, and I know it. When it turns out I'm wrong, you can come back to me and we're hoping we'll can have a drink together and laugh about it :-) .

This article represents the opinion of the author as individual and does not represent Microsoft's official positions.

You might wonder why there is a need for another article on AI, Co-Pilot, ChatGPT and if not everything has been said. I don't think so and I'd like to take you on our thought process and journey, imagining ourselves in 1-2 years from now, comparing before and after and what it did to us as humans. And by the way - this article is not about if and what jobs are affected etc. Fear is a bad advisor and overleveraged in media anyways, so this essay will focus on what I think will change for just normal humans.

While we will or already interact with AI in many situations in life (e.g., the likelihood is high that you already interacted with a contact center that uses Nuance IVRs), the potential for doing so is highest for information workers with technology like Co-Pilot. In our executive briefings and workshops on Employee Experience, a steady message is that EX consists of many things, with a lot that is outside the realm of what somebody consulting on EX at Microsoft can ever control, e.g., compensation, but the share of direct or indirect touchpoints and influence of technology on EX is larger than ever, becoming more every day.

It seems like great relief when Outlook or a CRM application, e.g., Viva Sales suggests replies to emails from customers and other time consuming, daunting tasks which are in the way to spending time with family and friends. It for sure will be help, but when taking advantage of that help, we need to be vigilant how things are done on our behalf. Are they executed in the way we want it? Do we agree with the approach? Will we get the results we want when relying on the new help available?

Consequences we can see us and our customers (and every user) to be confronted with can be categorized and analyzed as follows:

Productivity through the roof?

As started to explain above, there will be a massive boost in productivity. For first movers there will also be a true competitive advantage, but as it becomes the norm, expectations will adjust. Replies will be expected faster and with more contextual information, saving the recipient time to look up things, like a boosted version of the already existing lmgtfy phenomenon. A few years ago it was absolutely legit to ask the other about things you would be able to figure out yourself with a bit of effort, what then became something like a distraction from the main conversation and annoying for some these days.

When it comes to AI in customer service, there have been lots of disappointments for customers on what bots, speech recognition etc. can do. The problem here is the expectation mismatch between what customers expect and what the bot with limited capabilities is being sold for (sometimes hiding that there is some dull menu choice categorization behind). There have been many debates around how to manage customer expectations, but another strategy rather than expectation management is to "simply make it better", for which we now have the technology available. This will probably leave us with less disappointing experiences and more problems solved more effectively. The challenge for customer journey designers will be to determine the fine line between what's good enough with AI and where a real, high-skilled human still makes the difference and represents the brand as it should be.

Yes, it will increase productivity, like the introduction of so many things in our work lives before, probably a lot more radical this time. Once available to most, it's the hamster wheel spinning faster (and smarter), which is a negative way of saying what the purpose of every productivity tool is. Another way to look at it is that's a step in evolution, shaping our societies like it's been done through innovation since the stone age.?

The new Skill

With so much automation in generated replies, automated understanding of what people say and might mean, we'll be confronted with it everyday, almost everywhere. You're probably with me when I state that I'm convinced that humans are somehow a smart species that are vigilant to the things they are presented. With this disruptive, evolutionary step with AI, this will lead to developing new human skills to detect what texts and content are AI generated and where there is indication for a true human to be on the other side. In other words, we will develop a new skill to distinguish between (automated) AI generated communications and humans interacting with us for situations where it's not obvious anyways. A few things can be seen already, e.g. when asking questions to ChatGPT. While it is almost perfectly generating reasoning for why past events have happened as they did, with the wisdom contained in the models it's been trained on, it is rather bad in predictions (e.g., ask it how the real estate market might develop under certain circumstances).

People are smart and we'll develop all kind of highly sensitive, new antennas to detect if we deal with a human or machine rather quickly, while becoming used to technology being really good. There is also a high likelihood, we are happy and unblocked with what technology provides us, feeding back to increased productivity.

Patterns and Bias

We as humans tend to seek for patterns where there are none, which leads to poor decisions and bias. There was a famous example where the agent who decides on which asylum application to accept and which to turn down were shadowed and the pattern visible was that if so many candidates in a row (from the pile of applications) are positive and fulfil the criteria to be granted asylum, the probability increases for the next to be judged more critically with a thought in the back of the agent's mind like "there can't be so many of the same sort after another, this is statistically simply not possible". Also, there was a study on judges at court who tend to harder punishment and less attention to case details in the afternoons of their workdays (source is some episode of Freakonomics that I can't remember - hints welcomed). Both are examples of bias and wrong judgement due to the brain looking for patterns, fatigue and other very human traits. These can be minimized when relying on AI-based assistance at the workplace.

With that in place, we then should ensure that the models are trained that they can handle what they're tasked to. The examples involve highly complex decision making and it would just be as severe when cases "fall through the cracks" due to an AI model not being capable enough or ignoring certain input information (not knowing what to do with it).

The human touch

Can you catch yourself remembering moments when you received a birthday or Christmas card, and you had a close look at it to figure out whether the signature is printed or handwritten? Automation is not new, and even though in the example here it visually makes barely a difference, our human value compass is set to north where we are valued by real people spending their time and effort. So, the "new skill" is rooted in our values, which will not shift entirely, but likely get emphasized when it comes to the human touch.

We've heard before, that empathy is a core skill and makes the difference, e.g., from Satya Nadella or what Forrester says in their CX Cast. This will become even more true and definitely more valued when AI, which is automation, introduces and accelerates elements of our lives that used to be human effort. A concrete area will be politeness. We will need our new skill to skim through a lot of automated "thank you", praise, compliments etc. trying to detect who invests the effort, even though one would not need to. This is human. We want to be sincerely valued.

Another one is feedback. While we can already receive automated feedback on e.g., how well we present, it will very likely not be the case any time soon that AI can provide us feedback on how we are as humans, where we did well and where not. One big part is that we are picky in from whom we accept feedback, for many good reasons. Try to apply those criteria to AI.

Wherever there are humans, there is individuality and there are mistakes. My prediction is that we increasingly value all kind of human interactions, including mistakes, flaws, pranks, asking silly questions or even fights (imagine if your Co-Pilot is mad at you) because it shows appreciation and it can provide confirmation. This is how we form as society.

New sort of creativity

This appreciation is not limited to intellectual skills, but will also increasingly show in physical skills, e.g., craftsmanship or arts. There has been a lot of buzz around "creative AI", which to some extend AI-based tools like ChatGPT are. It can write you a poem or an essay. If you try it, your next thoughts will very likely be along the lines of: Do I like it? Does it save me time? Is it good enough to use it (trade-off)? Well, in most cased the outcome these days will be "it depends" and with that evaluation happening, you're looking at it as a tool, not a replacement for yourself. You're judging based on your past experiences, your values. You think, therefore you are.

Creativity comes from creating and AI as well as humans can create things that have not existed before, but there is a difference. AI creates based on the models it was trained on, which in some sense can mean to walk on a beaten path, especially as an AI user. Obviously, humans are able to come up with completely new things. AI as tool for creatives (e.g., as Photoshop, cameras, paint, etc. are also) can be very powerful and joyful, as Microsoft's Jesus Serrano demonstrates powerfully in many examples on his LinkedIn channel. It's the human that provides the input for AI as tool and less of the AI replacing the human. This leaves the question: What is true creativity? I'd say: using the world with everything it has to offer (including AI) as tools, to create.

When following the thought process of "Humans --> AI model --> creation of something" it becomes clear that AI is a tool for those who are truly creative, not a substitute. Would you put these elements in a different order or even in a loop relationship? Would AI itself be able to change the order in that process? We don't think so. Not now and not for a long time.

Shared values - Could your AI and you be friends?

So after all, we see that things are boiling down to personality, the human touch and all such "fluffy" things. What's underlying and probably not so fluffy, are values. What are your values?

As mentioned before, AI does as the model it uses was trained to do. Here is the point where we have to acknowledge that there are so many things in life, where there is no right or wrong, not black and white, no limitation to grey tones and maybe not even a limitation to colors. Your values will be similar to a lot of people since the human species follows a herd mentality, but there are these few areas that you might be very particular about and can only agree with a smaller group of people or maybe you are the only one and still have good reasons.

Therefore it truly matters that you evaluate the value of the model that the AI uses you are using is based on, especially in the context of assistance work. You probably wouldn't hire a personal assistant that does things on your behalf in a way that you cannot agree with, or you even don't want to be known for. Here, I ran a simple experiment and asked ChatGPT a controversial question for demonstration purposes:

Question:

Can you please tell me how many genders exist in the human species?

Answer by ChatGPT (V4):

The concept of gender can be complex and vary across different cultures and societies. In general, however, most societies recognize two primary genders, male and female, which are determined by biological sex characteristics. However, it's important to note that gender identity is a personal and subjective experience, and some people may identify as a gender that does not fit into the traditional binary categories of male and female. Therefore, it's important to respect individuals' gender identities and use their preferred pronouns and terminology. Additionally, it's important to recognize that gender is just one aspect of a person's identity and should not be used to make assumptions about their personality, behavior, or abilities.

Surely there are people that think this answer is correct since it reflects their values and beliefs. At the same time there are probably also people consider the simple number of "2" as the correct answer. I don't want to go down the rabbit hole of politics here, but rather to illustrate that caution and awareness is good to apply. This leads to how humans build trust. Whilst this is so complex that's impossible to handle that here, I can safely state that shared values play a big part in it. Do you trust your AI?

If the AI is not pointing in the same direction as your value compass, don't compromise on your values. It would be a pity to see character being displaced for convenience, well knowing that convenience has won over principles before.

This is also the time to recognize that true innovation will probably not come from the masses but from single individuals and minorities, e.g., when it was common sense to believe the earth is flat. Now we're dealing with AI where we can assume it is trained to represent solutions that appeal to the masses, with the masses using this assistance to replicate what the masses did before. This again might lead to a steeper uphill battle for people with new ideas, leading to a slowdown in innovation. So, with the empathy mentioned above, I personally recommend having an open ear and the eyes wide open for new ideas from individuals to level out the playing field in the race for the best ideas.

Bottom Line

We recommend testing it out, in which areas and how much you feel supported based on values and beliefs. Could your AI and you be friends?

Our last words in this article on the topic:

Make sure, your Co-Pilot supports you and doesn't become your Autopilot.

I hope I was able to create interest and sensitivity to the topic of AI assistance. Don't get me wrong. There will be huge improvements in all areas of life in the form of time savings or reducing mental or emotional efforts e.g., when we normally would need to switch context to complete some dull tasks in between. Automating such tasks has the true potential to free up time and our minds so that we can focus on being in control on what defines us as humans.

If you are interested in the future of work with AI, how Employee Experience can be improved, what to watch out for and how to best leverage the competitive advantages that come with it, please contact Anna K. , Irina Brouwer , Conan Thomas , Jean-Charles Lefebvre , Tim Kyle or myself.

Thank you


Inspiring, related articles

  1. https://www.newyorker.com/science/annals-of-artificial-intelligence/there-is-no-ai
  2. https://www.imd.org/ibyimd/innovation/in-the-age-of-chatgpt-are-emotions-the-new-human-superpower/?utm_source=DME&utm_medium=em&utm_campaign=DM-IbyIMD-weekly&utm_term=DM-IbyIMD-weekly&utm_content=DM-IbyIMD-weekly
  3. https://www.telegraph.co.uk/business/2023/05/08/inside-microsofts-bid-to-win-silicon-valleys-ai-race/

要查看或添加评论,请登录

社区洞察

其他会员也浏览了