Will AI Transform Retirement Funds?

AI is beginning to play a bigger role in trading. It’s been used on a large scale and in big investment firms for some time now, but of late we’ve seen more examples of private investors making use of the technology as well. Just last year, as you may recall, Royal Bank of Canada rolled out an AI trading platform for its global clients — basically giving those clients access to an algorithm that would assist their trading.

Moving forward, we expect this sort of AI application to become more mainstream. Private investors will increasingly gain access to tools that provide a variety of advantages, and even automate decisions in some cases. What is less clear at this stage however is how much of an impact AI will have on the management of long-term savings — such as those in individual retirement funds.

Aside from the general assumption that AI’s role in virtually all financial and investment practices is likely to grow, there are a few ways in which the technology could begin to transform retirement funds.

Improving Results

Naturally the primary intended benefit of augmenting retirement funds with AI would be to make those funds more likely to grow significantly over time. As one assessment of the idea of this technology in investing circles put it, AI won’t replace fund managers, but it does stand to improve their results. In this context, AI is viewed as a tool through which fund managers can analyze more data and gain greater insight, so as to make decisions more likely to result in strong performance. This means that while there won’t necessarily come a point at which retirement funds are turned over to AI with quick, dramatic improvement, we may well see gradually improving performance of such funds due to the technology’s influence.

Giving Investors Confidence

This point is baked into the previous one, but it’s also significant that improved performance would naturally breed greater confidence among investors. That is to say, people who may not prioritize putting money away for retirement may become more in line to do so if they begin to hear about retirement funds performing more efficiently. Thus, AI could not only stand to improve fund management, but also to lead more people to establish funds.

Communication with Investors

It’s a different function entirely, but the infusion of AI technology in retirement fund management also stands to streamline communication with the actual investors. A lot of people are somewhat uneasy with handing money over to fund managers and stepping back. AI can provide a way for those investors to stay in touch, so to speak. For example, an intelligent platform can tell investors about recent movement and decisions, and possibly even offer insights to support them. It may also provide specific guidance for investor decisions. Taking a standard Canadian retirement fund for instance, AI could communicate to investors how new contributions stand to affect deduction limits. It could also advise investors taking money out as to the tax implications of withdrawing funds from an RRSP. Advice like this isn’t always built into retirement fund arrangements, and plugging it in through automated means will go a long way toward making many investors more comfortable.

It will take some time for these and any related changes to take effect. Fund managers will need to onboard new technology, land AI talent, and adjust practices accordingly. In time though, we do expect to see AI affecting fund management, and hopefully boosting confidence in retirement savings in the process.

Meet the Athena Pathways Project Manager: Norma Sheane

Behind every successful project is an amazing team made up of individuals who provide significant impacts, resulting in successful outcomes. In this blog, we would like to spotlight one of the hardworking individuals who is responsible for the success of the Athena Pathways project: Norma Sheane, Project Manager of the Athena Pathways Project.

Norma is the main woman behind this project. Her tasks have included: monitoring all communication through the project mailbox, creating project procedures, content design, social media marketing, website layout and updates, manually making matches for the mentorship program, working with academic partners to determine applicable courses and eligible candidates for scholarships, working with Microsoft to offer their courses for free to BC women, working with companies who are hiring, managing the job board, coordinating all consortium members and reporting with the Digital Supercluster, and that’s just the tip of the iceberg! As with most project management roles, there are so many areas to focus on and pay close attention to in order for them to be successful. This role in particular has required a tremendous amount of communication and coordination between high level individuals with our academic partners and industry executives.

“A project like this requires a centralized person to act as the glue holding everything and Norma has been invaluable in that role,” said Steve Lowry, Executive Director of AInBC. Norma’s passion and dedication to the program have allowed it to thrive, and whilst she is not someone who would normally choose to be in the spotlight, we wanted our community to know who she is and the impacts she has had on its overall success.

Taking on this role was a risk for Norma. She came from a steady job in a unionized environment where she had benefits and security. Leaving that role came with the risk of not knowing the long term prospects of the role, not having benefits, and not having long term job security. She decided to take on the role of being the project manager of Athena Pathways anyways. When asked why she quoted, “While I’ve had great moments in every place I’ve worked, I wanted to make a difference. This project allows me to help hundreds of women and really gives me the opportunity to reframe the AI space in a positive, impactful way and to me that is worth the risk”.

Thanks to her hard work and dedication, Athena Pathways has now helped over 420 women enter or advance in the fields of artificial intelligence, data science, and machine learning. With the end goal of 500 women in sight, we are excited to have Norma continue as the project manager of the new Athena Pathways expansion project: Athena Digital Leaders. This new and exciting project will help both BC women and new immigrants find roles in tech by providing them with wage-subsidies, and tuition coverage to select courses.

Great work Norma!

The Entrepreneur's Guide to Fundraising

At first glance, raising funds for a startup seems organic - something you just do. But it turns out having a well defined process combining strategy, relationship building and technical considerations will drastically increase a founder's chance of success. In this conversation from the live webinar that took place on April 21, 2021, AInBC's Executive Director Steve Lowry is speaking with Russ Armstrong, VP Sales & Strategy of Boast.ai, about his proven fundraising process and his experiences raising $10M+ in private capital as Co-founder of Limelight Platform and working with Boast's startup client base to raise non-dilutive funding. In this recording, they discuss: Fundraising strategy - from bootstrapping to Series A Tactics for building and managing an investor pipeline Valuation multiples Equity, Debt & Grants

Watch the full recording here: https://www.youtube.com/watch?v=KliNPyUNwAM&t=1s

Proven Tips for Landing Top AI Talent

As published as a guest blog on HR Tech Group: https://www.hrtechgroup.com/articles/proven-tips-for-landing-top-ai-talent

Written by AInBC's Administrative Manager and Athena Pathways Project Lead: Norma Sheane

Artificial intelligence is a new and emerging field and there are some unique characteristics to look for when hiring for AI roles. Here’s a list of things to keep in mind along with specific examples to help you find and secure top talent.

1.  Define the role with specific terms 

It’s important to be specific about the AI skillset you need. 

For example, are you working with neural networks, and if so, do you need someone with experience with GANs? Until recently there were no degrees specifically in AI so there weren’t the obvious pathways into a career that more established disciplines have. Yes, many practitioners have studied machine learning in a Masters or PhD capacity, but you often encounter people skilled in this field who have come to it through a more generalized area of study like mathematics, computer science, data science, or physics.

If you need to hire someone who can speak French, stating that you need someone from Europe is not specific enough. You need to know what part of the AI universe your job candidates are coming from.

2.  Find lateral thinkers

The skillset for AI positions is different than for typical software development roles. 

In the case of developers, projects are typically well defined and structured, with clear milestones, start and endpoints, as well as little follow-up. AI projects often require exploratory and experimental work with unclear timelines. Plus, the nature of an AI system is that it will improve significantly once it gets exposed to first training data and then production data. It will constantly be improving and may require continual attention from the developer.

AI roles often require interaction and teamwork with a number of people in the organization because data may come from a number of different areas. This is also because the AI teams within a company are typically not large so practitioners need to handle many different parts of the project and the people in the company who interface with the system. 

Role requirements should include variety and critical thinking because the person in this role may need to "wear a lot of hats" and interface with many different parts of the organization. 

3.  Set your expectations on salaries

Here are general salary ranges in Canadian dollars for some typical roles:

• ML Researcher - $125,000-$350,000

• Data Scientist - $55,000-$190,000

• ML Engineer - $75,000-150,000

• Data Engineer - $55,000-130,000

In more senior positions, there is a big difference between good and great in terms of how far someone will be able to take your company.  Expect to pay competitive salaries for people who can easily pick up a job in global centres like San Francisco or Singapore.

4.  Work with agencies who know the area

If you went to France not knowing the language, would you want a foreign tour company to show you around or a French one that’s connected to the local businesses and knows the language? 

In AI, there are many places you can look to find different levels of talent but using a generic recruitment agency may not produce effective results as this area is highly specialized. Work with a recruiter with demonstrated knowledge and a network in this area - for example AInBC’s AI role placement service: https://www.ainbc.ai/placements

Job boards are typically effective however they can generate hundreds of applicants for a single role creating the challenge of filtering through and finding the small percentage of qualified candidates. Some VP’s say that 80% of the applications they’re receiving are not even relevant. For senior roles, the applicant pool may not produce even one individual who will be able to do the job at a high level. 

5.  Ensure ethical alignment

If you are building something that operates in an ethical gray zone - due to it being a rapidly developing area or otherwise - ensure you check the applicant's comfort level. For example, if you aim to make facial recognition technology, the person designing it will need to be comfortable with how your company will deploy it (i.e. for human health vs. for law enforcement). 

6.  For junior talent, ask around at universities

When looking for junior talent, contact universities as they can provide a number of opportunities to find talent. Professors can make recommendations of top students. You can sponsor hackathons and attend university hiring fairs and poster sessions. Also consider specialized talent programs that recruit students, programs like Insight Data Science.

7.  Check for hard skills

AI sits within data science, and since data science is a broad discipline which encompasses many skillsets, any number of people can call themselves a data scientist. But that doesn’t mean they know enough to do the job you have. 

Applicants should be able to demonstrate that they can work with the technologies needed to commercialize the product. Academics may not have needed to code production-ready systems and may have relied on others to do coding work. 

Applicants should be given a scenario to solve, to demonstrate they can be creative and think outside the box. They should also be able to demonstrate that they stay up to date on the latest developments in AI since the pace of change in the industry is so rapid. 

In summary

These tips can help you navigate the new realm of hiring AI talent. It also helps if you know someone who’s an AI expert. Reach out to them for a virtual coffee and pick their brain. It will be well worth it to get your data straight and your machines set on “smart”! 

Do you want to know more about Artificial Intelligence in BC? Check out the AInBC and Athena Pathways websites, or contact info@ainbc.ai for more information.

Head to Head: 4 Canadian companies joining Google's Voice Accelerator - who has the best business?

Four Canadian AI startups have been awarded almost half the spots in the new Google for Startups: Voice AI program. This is fantastic to see. They’re also fairly representative of Canada on the whole, coming from Vancouver, Toronto and Halifax.

I wanted to first recognize these impressive accomplishments, and then look at these voice projects as businesses - comparing them side by side in the first of a new “Head to Head” series we’ll run to unpack some of the fundamentals of data-driven businesses. Hopefully the competitive aspect will make for fun reading.

I’ll first outline what each company does and then compare them across four parameters for data driven businesses.

SiMBi teaches children to read by guiding them word by word through stories.

Babbly helps parents develop their babies’ language skills by giving feedback on early verbal sounds.

Talkatoo is a dictation note-taking platform for health professionals - primarily veterinarians.

ConversationalHEALTH is digital agent enabling patients to better interact with the health system by tracking prescriptions, appointments and other aspect of their care.

Pain Point

If we apply the traditional Aspirin vs. vitamin comparison, to see if a product alleviates a customer’s major pain (an Aspirin) or merely makes them stronger and smarter (a vitamin), SiMBi appears to fall more on the vitamin side of the line. Since it is designed for the mass market, it has the potential to help millions of children read better - a very worthy goal - however parents and teachers may question if this is something they need to do this job or merely a nice to have addition for variety. Babbly is similarly a tool that can be used by all parents, but given that it connects parents to speech pathologists in cases where there child may be developing slowly, it would no doubt be viewed as an Aspirin for the subset of the market that feels their child is lagging behind. Talkatoo has some users who swear by its product, but it does feel more like a vitamin, comparable to an Alexa skill company in the legal space called Tali, that ultimately ceased operations. Talking to a computer throughout the day seems like the logical evolution beyond typing, but it may be early in either the user experience or user adoption of the technology to build a large business, particularly given that dictation is widely available out of the box in Microsoft Word. ConversationalHEALTH meanwhile, distinctly offers an Aspirin - quite literally as the first example on their homepage shows. Medical services are both highly complex and highly important and patients frequently experience communication issues with their healthcare providers as well as errors in self treatment.

Market Size

All of these companies are tackling massive markets. SiMBi and Babbly have products that could theoretically be used by every child on the planet. Applying a conservative churn rate of 10%, the lifetime value (LTV) of a SiMBi introductory customer is $500 (parents pay $99/year). It’s not evident what their cost of customer acquisition is, but it may be difficult to get a return on acquisition costs if the majority of customers are parents rather than higher-paying schools. Babbly does not have a SaaS model, so they may also find monetizing difficult if revenue from training courses and speech therapy sessions is highly episodic. Talkatoo appears to be deploying the compelling strategy of targeting a large market (everyone who could potentially dictate their work) while beginning with an approachable beachhead (veterinarians) and then expanding to other areas with custom-built lexicons. They apparently have a higher LTV of $11,500, but they will likely face a lot of competition in other verticals and they may only be able to claim this first vertical they are pursuing as their own. With ConversationalHEALTH, the costs of healthcare inefficiency worldwide are huge, not to mention malpractice costs, but this is likely to be one of the most contested areas of all AI activity in the coming years so their success rests on being able to effectively differentiate.

Data Advantage

The consensus view of AI businesses is that good data beats a good algorithm every time, and as a result, proprietary datasets typically comprise a high proportion of the value in AI startups. SiMBi and Babbly really shine on this measure because children’s speech patterns are highly unique. SiMBi has come up with a clever way to diversify its dataset by having children read to each other around the globe. Meanwhile, I can think of no way a company could obtain the training data that Babbly has at scale without copying their business. Talkatoo also has a proprietary dataset that they can improve on incrementally with every new veterinarian who uses their tool. ConversationalHEALTH doesn’t initially claim anything particularly unique in their dataset that other health tech companies wouldn’t be able to develop, so are likely in a footrace to gather one of the largest datasets in lieu of differentiation. The larger an AI model is, the better it can address edge cases - which of course can be critical when human health is on the line.

Other Competitive Advantages

All four startups are pursuing pro-social aims, for example SiMBi’s claim to double reading fluency in three months. This will help them all attract talent and likely funding. In terms of unique positioning, SiMBi’s assisted reading algorithm is probably not difficult to replicate, but they have the potential to create network effects with their users sharing recordings with others around the world. Babbly, on the other hand, would appear to have a highly differentiated algorithm that can parse baby sounds for signs of language fluency. As stated, Talkatoo’s potential lies in its ability to dominate a niche, and perhaps doesn’t have a lot of other strong differentiators yet. ConversationalHEALTH is in the a gold rush of companies developing AI for health diagnosis, treatment and management, but unlike a gold rush, the vast majority of the profits will likely end up in the hands of just a few dominant prospectors. In order for ConversationalHEALTH to be one of them, they will need to develop compelling differentiation, whether through network effects or otherwise.

Ultimately, I’m not aiming to declare a winner, but I am interested to hear what readers think about the companies’ prospects. The one undisputed winner of the publicity Google endorsements provide is Canada’s tech ecosystem. We all benefit from reinforcing the country’s expertise in AI.

Funding goes up, funded count goes down

Stanford’s Institute for Human-Centered Artificial Intelligence just released a report showing that funding for AI companies and projects increased 4.5 times between 2019 and 2020, but curiously the number of AI startups receiving funding went down for the third year in a row.

Does this suggest we are “moving from pure research and exploratory small startups to industrial-stage companies”?

Almost Everyone Has Baumol’s Cost “Disease”

But it’s not as bad as it seems.

Baumol’s Cost Disease has been around since the 60’s and everyone who’s been to a hospital or a university has experienced it. Fortunately it doesn’t hurt your health, but instead it hurts the healthcare system, and it’s something that Artificial Intelligence can treat.

The Baumol Effect, as it’s also called, was defined by a pair of economists as the rise in costs that happens in an industry that falls behind other industries in terms of productivity gains, provided workers can switch between industries.

For example, the healthcare industry has resisted technological advancements that would reduce the labour - i.e. the time of doctors and nurses - needed to treat patients. At the same time other industries have deployed technology that makes labour far more efficient and leads to increases in the value of the labour and thus salaries. Think mass media, agriculture and manufacturing.

In order for hospitals to keep their staff from retraining as television producers, they need to continually increase staff salaries relative to consumer purchasing power even though the staff members perform substantially the same jobs they did 50 years ago. These costs ultimately must be borne by the consumers of health services, and so we’ve now seen health costs balloon to 1/6th of GDP in the US. Meanwhile video content, food and shoes have gotten far cheaper over time.

Marc Andreesen and Vijay Pande released an excellent A16Z podcast on this where they outline the situations where AI is highly effective at reducing costs. From their conversation we can see that AI entrepreneurs and investors will do well to focus on businesses that use AI where:

1. There is a large amount of information that is time consuming for humans to process but easy for machines - e.g. by supporting billing and triage in hospitals, technology can free doctors up to do more important work.

2. The free market is at play - e.g. the technology in Lasik eye surgery has evolved far more rapidly than heart surgery in part because the user pays out of pocket.

3. An AI can be taught a subject once and then the knowledge can be widely scaled, as opposed to needing a person to teach the subject to others repeatedly.

4. A service can be converted into a good - e.g. digital technology can take the performance of a musician (a costly service) and replicate it into millions of playback feeds instantly for pennies.

We can see these processes at work in two exciting Canadian startups. First in healthcare, Vancouver based Satisfai Health is using computer vision to diagnose gastrointestinal diseases in real time. Images of polyps contain a huge amount of information and AI is becoming more accurate than humans at finding malignancies. The result is that Satisfai’s machine learning-enabled optical biopsies avoid the need for many highly expensive and invasive physical biopsies. In addition, their system helps lower drug costs by enabling pharmaceutical companies to terminate trials of nonperforming drugs earlier in their development and focus on more compelling candidates.

The second startup, Korbit out of Montreal, is an education company.

SOURCE: BUREAU OF LABOR STATISTICS

The real shocker in this Consumer Price Index comparison graph is higher education. Andreesen and Pande point out that the cost of a four year college degree is on a path to reach $1,000,000. A vexing trend, especially compared to the cost of televisions, which they say before long will be the size of your wall and cost $100. Tech pundit and NYU professor, Scott Galloway, has written much about the skyrocketing cost of higher ed, including in his latest book, Post Corona, where he explains why this space is on the cusp of massive disruption. Before long, he expects many software developers to go to school at Google and Apple and receive an education similar to at a university. It’s not that universities will go away but instead the hope is these venerable institutions will adapt their service delivery model to provide education to more people at a lower cost.

Korbit is leveraging the training power of machine learning with their AI tutors, which customize educational content for every student based on the pace of their learning. They are also converting a service (teaching) into a good (customized learning on demand). It would be cost prohibitive to have a teacher teach every student a unique learning plan one-to-one but the incremental cost of each new tutor “feed” is negligible. Korbit is backed by Khosla Ventures and Real Ventures.

What other industries are primed for AI led disruption - i.e. where automation would reconfigure the labour inputs to lower costs and improve user outcomes?

Start Here

From here on, things will be different.

Different and better, I want to say. With the spirit of a spring festival, the first half of 2021 will see many absent hopes return as vaccines and antibodies proliferate. There’s now a feeling of renewal in world leadership as climate initiatives are again prioritized in the United States. Inequality is on everyone’s mind, though it remains to be seen whether it will be decentralized finance and stock trading mobs or well organized and intentioned community groups that make the first truly meaningful openings out of all the cracks in the facade.

In technology, where we work, several astounding leaps are being made in health treatment, and a desire for innovation has set into higher education, which will hopefully stop the spirals of rising costs in these fields. Both artificial intelligence and startups are playing leading roles in these changes. In AInBC’s backyard, for instance, we see startups like AbCellera, Variational AI, 1QBit and Emtelligent working nonstop to fight COVID and make health treatment more efficient.

As always, when it comes to AI, AInBC wants to be part of the solution. We unfailingly support pro-social AI initiatives and innovators and we are now reinvesting in the area of capital raising for AI startups. Coming on the heels of our recent Launchpad competition, where we invited eight of BC’s most promising AI startups to pitch investors around the world for seed and growth capital, we are launching this blog to investigate, profile and debate issues in raising capital for AI startups. We seek to engage both founders and funders who see AI as a transformative technology for good.

AInBC is turning two this spring, and as icing on the proverbial cake, we also want to announce a new advisory program to train AI startups and help them raise capital faster. This is essentially a grant provided from our members through their membership dues to you, the startup community! If you feel you could use a few Zoom calls worth of advice on fundraising, technology, IP, new customer development or operations from successful founders and technology leaders in our network, no strings attached, then reach out to us on our contact page or via LinkedIn and we will do our best to connect you with someone who can share actionable insights to help you grow.

AI today is a colourful, burgeoning and amply-hyped space, but at the end of the day when the dust settles, it’s the positivity and drive of the people working here that imbue the technology with so much promise. We’re here to help those people by building an AI community with strong ties between its many members.