How much should this AI Chatbot Writer job pay?

Hybrid jobs are all the rage currently and are some of the top paying jobs in the market right now. If you’ve got soft skills, business acumen, and technical skills, you’ve got the ticket to a high paying job.

Hybrid roles are super interesting to follow because they are so new. Their descriptions and responsibilities differ from one organization to another. This is particularly the case with AI interaction designers, a emerging job category I’m paying a lot of attention to lately (in part because I’m slightly obsessed with chatbot design as of late.) Diane Kim, who designs the friendly virtual assistant bot at x.ai, summed up this emerging field in her interview with Wendy and Wade, a career advising chatbot:

“The fact that AI Interaction Design is so new gives me the freedom to be experimental. I also have the unique opportunity to be part of defining an entirely new field. This is actually both what is most exciting and most challenging about my job…But it’s challenging because none of us really know what this is yet — we’re all figuring it out together. It’s really different from, say, being a recent grad in your typical UX role for a visual interface, with decades of research and best practices to follow. We don’t have the same industry standards or guidelines yet for conversational design, but the fun part is figuring them out as we go.”

So it’s within that context that I examined this AI chatbot writer role from JustAnswers.

Chatbotjob Chatbotjob

The skill requirements on this role are massive. Let’s break it down.

  • You need quantitiatve and qualitative skills
  • You need to be a seriously good at writing (perfect tone!)
  • You need to understand Sales (identify (and contribute to?) revenue opps!)
  • You need be an experimenter – test and retest
  • You need mad research skills
  • You need the collaboration skills to work with diverse teams
  • You need to understand user experience
  • You need to dive into professional fields that requires years AND be required to anticipate which quesitons users would ask AND write the answers.

This is one hell of a robust skill set. That last ask – expert with diving into deep professional fields like medicine and law – really threw me off. Who is this person? And will you pay them a shit ton of money for this expertise and skill set?

It’s likely this job is like most job postings: crammed with all the ideal things. There is probably flexibility – an applicant doesn’t have to have all those things.

I’m curious about how much this role pays because writing is an underpaid profession. Some managers who don’t write assume it’s easy – after all they write emails and reports! Copy is everywhere and people assume it’s easy to produce. Thoughtful copy – the kind that strikes the perfect tone! – takes time and creativity to produce. People in quantitative fields tend to overlook that.

But bad writing, especially in AI conversation design, leads to awkward interactions with the product. For example this was my recent convo with a new recruiting bot Robo Recruiter:

If writing is underpaid but AI is a hot hot hot field, how much should we be paying our AI chatbot writers?

I’m crowdsourcing your answers below in the comments: how much do you think this job pays? Do you think it pays as much as a machine learning engineer? As a product manager?

Write your answer below.

Then see what Paysa pegs the going salary rate in San Francisco.

 

How to learn about ML/AI if you don’t have tech skills

Art by AI

I’m a liberal arts grad. I love words and language. I teach soft skills. Qualitative data is my jam. I’m also obsessed with machine learning (ML) and artificial intelligence (AI).

In 2015 I tumbled down the AI rabbit hole after discovering a long read on the fabulous site Wait But Why. The site explains complex ideas paired with hilarious stick figures. The two part series on AI, The Artificial Intelligence Revolution, was my gateway article to the world of AI, and later ML as part of AI.

So far my self-directed learning journey has only included reading about AI and writing about its affect on hiring and the future of work. I can’t code in Python (with zero plans to do anything with R). My data background includes data analytics, cleaning data, and putting it into Tableau but nothing close to data scientist. I also have no interest in going that far professionally. As a non-tech person trying to access ML/AI, it’s been a challenge to figure out where I fit in. I’ve uncharacteristically avoided meetup groups or conferences on the subject since I don’t have the tech skills.

Not me.

Last month I changed that. I got tired of reading. I wanted idea exchanges. So I attended a ML/Al unconference in PDX. And hot damn I found my people!

An unconference is the opposite of the standard conference setup. Instead of corporate-sponsored keynotes paired with bland chicken and an abundance of shy speakers who read PowerPoints, the participants chose the content. We pitched and voted on what they wanted to talk about. The result was facilitated conversations about subjects we were curious about and a format that flowed. It was the ideal setup for idea exchange and learning. If you’re conference weary an unconference will restore your faith in professional development.

Many people at the unconference were data scientists or computer scientists, and some working on ML projects. A few were students or job seekers. I met one other person who is like me, a communications expert without a technical background who works for a machine learning platform, BigML (and they’re doing rad stuff).

In our sessions we covered a roving range of topics about ML/AI: novel data sets, making AI more accessible to the masses, establishing trust with users, data security, AI decision making re: self-driving cars and the Arizona accident, becoming a data scientist and machine learning engineer, the future of companies and jobs (my pitch!), learning ML/AI as a new person (do you learn the math, the code, or find a project first? plenty of debate on this!), and plenty more side conversations that spilled out of the main sessions.

After years of consuming content about ML/AI I was thrilled to have in-depth discussions about it. (sidenote: the majority of my professional network and peer network are not tech people – their eyes glaze over a bit when I start talking about it.)

As an non-tech outsider it’s a bit intimidating to participate in such a cutting-edge tech space. I think ML/AI people forget that at times. One of the guys I met at the conference noted that when you’re an expert it’s hard to remember how hard it is for others to start in your field. I’ll add that this goes double if you’re in a quant and code heavy field like machine learning. Luckily most everyone at the unconference made it easy to participate (as did the unconference format).

My main takeaway though is that you don’t need to be a software engineer, data science expert, or code wizard to understand ML/AI.

So for all the people who are curious about ML/AI but don’t know how to start engaging in these communities, here’s how. 

Learn the basics: Know the difference between machine learning and AI; understand the difference between Artificial Narrow Intelligence, Artificial General Intelligence, and Artificial Super Intelligence; understand the basics of data science. There are no shortage of intro articles and videos on the subject (two examples below).

Here’s a helpful Quora answer about the differences between a data scientist and a machine learning engineer. 

Prior to the uconference I was slightly worried I’d be left out of the conversation if it turned to technical. I prepared by returning to a set a YouTube videos I’d skimmed a while back: Fun and Easy Machine Learning. The YouTube list animates over 15 models to better understand machine learning.

Ignore the math and coding right now: Unless you want to become a data scientist or machine learning engineer, ignore it. You don’t need it to understand the basics or to explore products or impacts of ML/AI. For example, the Fun and Easy Machine Learning series sometimes dives into the math behind the models. Treat it as you would a foreign language; when you don’t the meaning keep moving forward and focus on what you do understand. Fill in the blanks later.

Read everything about ML/AI in the area you’re interested in. ML/AI for non tech people is a huge field. So narrow it down. Start with general articles about artificial intelligence and learn about it’s expected impact. The World Economic Forum has good articles with a global perspective. For business impacts, check out this history of ML/AI technology by industry/verticals. Then head over to CB Insights to study ML/AI companies (and subscribe to their newsletter as they’re cutting edge everything). Then pick an industry that interests you. Either one that you work in or one that you want to work in. Read everything you can about how machine learning is affecting that industry (it’s affecting all of them – right now finance, healthcare, and insurance are some of the industries talked about the most.) Explore products and platforms in that industry that use ML/AI. Read case studies. I study the future of work. So I read everything I can about ML/AI and it’s affect on workers and organizations: McKinsey, AXIOS, MIT, plus I play with HR Tech.

Avoid the hype. It’s easy to get caught up in the shiny promised of AI. Instead, pay attention to counter narratives, often published outside of the tech reporting ecosystem. Find the counter narrative about AI in your field. I read the amazing research and work by Audrey Waters at Hack Education for a counter narrative to AI edtech hype. Explore bias in ML/AI. Understand how AI isn’t neutral and that gender and race bias is coded into AI systems. Weapons of Math Destruction is an excellent book (and 99% Design has a good podcast on it). We need diverse perspectives and people in ML/AI fields to fight these bias, and non-technical people are part of that fight. 

Go to an event and talk to people: This is the intimidating part. But get over it, embrace the awkwardness, and commit to asking curious questions. Remind yourself of the things that you know. Write down the things that you want to learn. Talk to people until you get the answers to your questions. Ask people how they got into their work, what impact they’re having, and how they’d explain their work to a non tech person. Tell them you’re curious. Some people will just talk at you. Others will teach you. Keep in touch with the people who teach you and simply move on from the ones who talk at you.

Get a project: This builds on not worrying about the math and coding. Instead, get a project. What problem do you want to solve? What problem does your organization need to solve? What data is available? What data is missing? How could ML/AI solve your problem? Starting there will help you lead you in the right direction. You might not have an answer right away. That’s ok. It make take a while to solve it. But that’s the point. You’re learning. Ambiguity is part of the process. So ask around your workplace. Visit the data science or computer science team in your organization (assuming you have one). Find a data scientist in your network or at ML/AI events and ask them how they’d solve your problem. Ask them to break it down. Ask a computer science student what they think.

Start with curiosity, ignore the part about not having a technical background, and see where it takes you.

Navigating AI in the Job Hunt

Just dropping this Guardian article off here: ‘Dehumanising, impenetrable, frustrating’: the grim reality of job hunting in the age of AI

It features plenty of questions we should all be asking about AI in the job search. It also centers the discussion on the maddening experience of searching for work when AI is your evaluator and the gate keeper to getting hired. It’s ironic that organizations want more employees with soft skills yet the recruiting experience is transforming into a less human process. On top of that we’re outsourcing the ability to identify the relevant soft skills to technology that still isn’t very good at them.

This shift has already radically changed the way that many people interact with prospective employers. The standardised CV format allowed jobseekers to be evaluated by multiple firms with a single approach. Now jobseekers are forced to prepare for whatever format the company has chosen. The burden has been shifted from employer to jobseeker – a familiar feature of the gig economy era – and along with it the ability of jobseekers to get feedback or insight into the decision-making process. The role of human interaction in hiring has decreased, making an already difficult process deeply alienating.

Beyond the often bewildering and dehumanising experience lurk the concerns that attend automation and AI, which draws on data that’s often been shaped by inequality. If you suspect you’ve been discriminated against by an algorithm, what recourse do you have? How prone are those formulas to bias, and how do the multitude of third-party companies that develop and license this software deal with the personal data of applicants? And is it inevitable that non-traditional or poorer candidates, or those who struggle with new technology, will be excluded from the process?

Job seekers will be battling the robots on two sides: in the recruiting process and as they advance in their careers. It’s not going to get any easier.

Are employers telling candidates AI is evaluating them?

There’s a stand out line from a recent INC article on how AI is changing the hiring process. In the post, AI Is Now Analyzing Candidates’ Facial Expressions During Video Job Interviewsthe journalist asks:

“Are job candidates told that their facial expressions will be analyzed by algorithm?”

It’s a basic question that needs more examining as new technologies that use AI to screen candidates become more mainstream in the hiring process. The product in question here is Hirevue, a video interview platform that uses machine learning to make predictive assessments about a candidate’s future performance. It’s received over $93 million in funding and is used by a variety of organizations like Unilever, Goldman Sachs, Atlanta Public Schools, and BYU.

HireVue is one of the most high profile technologies in the HR Tech space. They’re using technology that enables recruiters to hire more efficiently. But the technology fundamentally changes the way candidates interact with employers and how they are evaluated. A journalist over at Business Insider tried the software and describes the process:

HireVue uses a combination of proprietary voice recognition software and licensed facial recognition software in tandem with a ranking algorithm to determine which candidates most resemble the ideal candidate. The ideal candidate is a composite of traits triggered by body language, tone, and key words gathered from analyses of the existing best members of a particular role.

After the algorithm lets the recruiter know which candidates are at the top of the heap, the recruiter can then choose to spend more time going through the answers of these particular applicants and determine who should move onto the next round, usually for an in-person interview.

The journalist also reported how awkward the experience is. You’re not interacting with anyone during the experience. Instead you’re staring at your own face. And it’s not just journalists who feel this way. For a good chuckle, take a look at the feedback on a HireVue experience on Reddit:

Hirevue interviews are awkward as well. from jobs

Should I send a thank you letter after a Hirevue interview? from jobs

Was Goldman Sach’s HireVue interview really awkward, or is it just me? from cscareerquestions

Interestingly none of these posts talk about being evaluated by AI. A quick look through company tutorials on how to use HireVue doesn’t say anything about AI making judgements about your microexpressions and voice.

Obviously this isn’t a representative sample. But companies have a responsibility to tell candidates how they’re being evaluated. And candidates need to ask tougher questions about the evaluation process so they can prepare and adapt accordingly.

And for job seekers who are navigating this impersonal world of HR Tech, here’s some handy advice from the INC article:

“For job candidates, knowing your emotions will be read, it’s a good reason not to apply for any job or to any company you’re not genuinely enthusiastic about. Or it may be a good reason to brush up on your acting skills.”

If you’re curious, here’s how HireVue works:

 

So about that job offer at Facebook

Corporate surveillance is all the rage among the top tech companies according to this Guardian article, How Silicon Valley keeps a lid on leakers:

For low-paid contractors who do the grunt work for big tech companies, the incentive to keep silent is more stick than carrot. What they lack in stock options and a sense of corporate tribalism, they make up for in fear of losing their jobs. One European Facebook content moderator signed a contract, seen by the Guardian, which granted the company the right to monitor and record his social media activities, including his personal Facebook account, as well as emails, phone calls and internet use. He also agreed to random personal searches of his belongings including bags, briefcases and car while on company premises. Refusal to allow such searches would be treated as gross misconduct.

There are some truly shitty practices happening at top technology companies like Facebook and Google. The paranoia is so bad in some companies that “some employees switch their phones off or hide them out of fear that their location is being tracked.”

So how does a job seeker know to avoid companies that treat their employers like this? And does it even matter because the long term benefits of getting Facebook or Google on your resume and working on cutting edge projects outweigh the risks of daily corporate surveillance? (yes, it should matter, but try telling that to a new graduate)

Maybe these practices are more of a reflection on just how comfortable we seem to be getting with corporate surveillance in our professional and personal lives.

Interview with a chatbot part 2

These past weeks I’ve been deep into the #HRTech world, tweeting frequently into the void, trying to learn more about increasingly opaque data used in smart HR platforms. Throughout the process I’m documenting the variety of hiring technology on the market, from smart platforms to machine learning for automated resume screening to virtual assistants. Along the way I’ve stumbled on loads of chatbots trying to claim a place for themselves in the hiring process. I’ve got a bit of a crush on chatbot technology so I’ve been trying them out. Two weeks ago I pined for Mya but settled on an interview using TalkPush. Today I found Paradox.ai and gave it a go.

Once again, the intro starts easily enough:

Then I was immediately asked contact details. Mind you, I’m just browsing here, not actually ready to apply. I don’t know if it’s me or what but I’m a bit irritated each time I’m asked for contact details right away (side note: I signed away my LinkedIn data for Wade&Wendy access only to be told post-data exchange that I’m on a waitlist, so maybe I’m just tired of having to give up data to engage). But this is all pretend anyway, so I gave them my phone number and then we moved on to my interests.

I thought here that we might talk about what positions are available but the onus was put on me to define what I want. I actually wasn’t sure what to answer. I like that it’s framed that way but wonder how other job seekers perform when asked this question. Admittedly it caught me off guard and I wasn’t sure what to write. I had to think about it which then sent me down a mini-spiral wondering if they evaluated me on how long it took for me to answer.

Moving along:

This is where it got interesting. They’ll find my profile (and I wonder if they’ll find my other social profiles) for me, so I don’t have to submit anything.

I think there was a hiccup when I shared my non-existent most recent role as the interview ended abruptly. I can’t tell if it’s because a. it’s a bot. b. I’m not a fit. or c. this wasn’t an interview.

Then onto the questions from me. The Q&A started a bit rocky but got better:

So throughout this whole process I wondered: how do you know the difference between a bot that helps you explore job opportunities and one that evaluates you? I misunderstood this bot from the beginning.

I went into it thinking the bot was helping me explore options at the company. But it quickly moves into interview territory by asking my experience level. Then it forces me to automatically apply to the position(s?) we discussed as they pull my LinkedIn profile. What if I hadn’t updated my LinkedIn?

Also, what if I’ve already provided my real name and contact, but wasn’t prepared to discuss my experience, what do I do? If I abandon the convo, and return, how does that affect my evaluation? Am I more desirable because I’m returning? Or am I penalized because I couldn’t answer the prior questions?

I’m so curious what happens on the backend when a recruiter receives the data.

While this experience is certainly efficient it’s hard to get a feel for company culture during these interactions. I was generally curious about the companies that they partner with but didn’t get traction there. Asking about the workplace and getting a canned response about “best talent” and “superstars” doesn’t offer much. If Olivia instead shared a video from the team, or a blog post about a day in the life of a marketer at Paradox, or even a personalized message from the founder that wasn’t full of “superstar” startup speak, it’d instantly provide more value. It’d at least add a personal touch.

Interacting with bots has me wondering how we define candidate engagement within the context of chatbots. Olivia engaged with me but she wasn’t engaging (though she was definitely better than previous bots I’ve engaged with). When the novelty of interacting with recruiting bots wears off (it’s still so very new), I wonder how candidates will view the experience.  If there’s a war for talent, how do you expect someone to chose your company if you can’t show off goods? Do bots play a role in wooing candidates? Or are they just there to expedite the hiring process for HR?

And if candidates are expected to show their soft skills, how do employers expect to identify them when the majority of HR tech aims to take humans out of the selection process?

Start upskilling for AI now

In 2017, roughly 70,000 postings requested AI skills in the U.S., according to our analysis of job postings. That’s a significant change, amounting to growth of 252% compared to 2010. Burning Glass also found that demand for AI skills is now showing up in a wide range of industries including retail, health care, include finance and insurance, manufacturing, information and professional services, technical services, and science/research. – Burning Glass Technologies

I’ve been seeing AI skills pop up in random job posts. I’ve wondered if it’s part of a bigger trend. It’s hard to get perspective since I’m not in the job market. Amazon leads the hiring for AI skills by a mile but GM, Accenture and Deloitte are also investing heavily. The most in-demand AI skills:

software developer/engineer, data scientist, data mining/data analyst, data engineer, computer systems engineer/architect, medical secretary, systems analyst, product manager and business management analyst.

Medical secretary threw me for a loop. Maybe because they’re working with new AI medical technology? Regardless it’s time to upskill.

 

AI for the doctor’s office

SmartExam acts as a virtual physician’s assistant – an automated medical resident, if you will – that enables primary care providers to deliver efficient remote care while cutting costs and improving outcomes… The intelligent software dynamically interviews patients, using answers to garner more information and support providers in the care delivery process… SmartExam lets providers achieve as much, or more, in a two-minute virtual patient visit as the 20 minutes of provider time needed for an office visit, the company said… “It allows clinicians to operate at the tops of their licenses,” said Constantini. “They can focus on what they do best — diagnosis and treatment.” – Bright.MD raises another $8M for “virtual physician’s assistant” SmartExam

I wonder if current medical students are taught how to integrate AI software into their training.

Hiring practices are about to get even more opaque

All that advice about plugging keywords into your resume to make sure it passes the ATS systems is about to be useless. Here’s an excerpt from AI for Recruiting: A Definitive Guide to for HR Professionals by Ideal.com, a AI-powered resume screening and candidate tracking solution for busy recruiters.

Intelligent screening software automates resume screening by using AI (i.e., machine learning) on your existing resume database. The software learns which candidates moved on to become successful and unsuccessful employees based on their performance, tenure, and turnover rates. Specifically, it learns what existing employees’ experience, skills, and other qualities are and applies this knowledge to new applicants in order to automatically rank, grade, and shortlist the strongest candidates.The software can also enrich candidates’ resumes by using public data sources about their prior employers as well as their public social media profiles.

Now for all the questions: What are the “other qualities” that they measure? How much weight do they give to experience vs. skills? How much data does a company need to use these algorithms effectively? How does a company without loads of data use this technology? Who decides which data to use? Who reviews the training data for accuracy and bias – the company or the vendor? How does this company avoid bias, especially if people who advance are all white men (due to unconscious bias in the promotion process)? What data points are most valuable on candidates social profiles? Which social profiles are they pulling from? Are personal websites included? Which companies are using this technology? Are candidates without publicly available social media data scored lower? Of the companies using these technologies, who’s responsible for asking the questions above?

This technology gives a whole new meaning to submitting your resume into a black hole.

Harsh words, harsher realities

“You’re out of time. If you can’t already write a piece of code to find the longest palindrome in a string, you probably won’t be able to do so before the automation revolution deals a body blow to your banking job sometime around 2022. Cathy Bessant, the chief technology and operations officer at Bank of America, said as much in conversation with Bloomberg last month. If you’re a bank employee who’s technologically illiterate, Bessant said it’s no good rushing to do a few coding courses on the side. You’re too late: things are moving too fast. “The kind of skills that we’ll need have to be taught beginning at a much earlier age,” said Bessant. “Whether you can train the same worker at the same time you’re changing their job remains to be seen.” – Can’t Code? The only other thing that will save your job

Immediate thoughts: 

Does this include executives and leadership?

Are they doing any work to train their best and brightest in these skills?

Where will these bankers go now?

Does it even matter since this is the new reality:

“Huy Nguyen Trieu, the former head of macro structuring at Citi, told us he knows of a team of just four algorithmic traders who now manage 70% of the trades that were done by 140 people in 2010″

Luckily there’s a sliver of hope for bankers in form of soft skills:

“Not for nothing has Goldman Sachs president David Solomon been extolling the virtues of a well-rounded education that incorporates public speaking and communication. Just as banks need geeks, they’ll also need exceptionally charismatic individuals to act as the face of the new automated reality.”

Maybe I should launch a new workshop as part of my power skills series: How to Charm the Pants Off of Your Audience and Save Your Job