The Future of AI and Its Ethics: An Interview with Daniel Shapiro of Lemay.ai

Who Is Daniel Shapiro

Daniel Shapiro PhD is the co-founder and Chief Technology Officer of Lemay.ai. With extensive experience in top management and consulting roles, Daniel has dedicated his career to advancing the field of machine learning and leading projects with major tech and defense companies in Canada. In addition to that, Shapiro is also the co-founder and CTO at AuditMap.ai, and serves as an advisor for PH360 and biological neural networks (Nuraleve).

The past few months we have witnessed impressive breakthroughs in the field of AI. Language model ChatGPT has taken the world by storm, while the creation of MidJourney has spark numerous debates surrounding AI and art. Moreover, tech giants such as Amazon and Google are on the verge of launching their artificial intelligence solutions.

While it's an exciting time filled with endless possibilities, it also raises concerns and uncertainty. With the potential impact AI could have on the job market and how it may evolve in the future, it's crucial to have a deeper understanding of the technology.

In this interview, we talked to Daniel Shapiro, the CTO of Lemay.ai, a leading AI consulting firm, having to worked with some biggest names in technology such as NVIDIA and Amazon, as well as naval defense and aerospace agencies. Our conversation delved into the vast potential of AI, its risks, dangers and the ethics of its development.

Spotlight: AI and machine learning is rapidly growing, with art apps like Midjourney and language models like ChatGPT. How far has this technology developed?

Daniel Shapiro: It's both humbling and exciting to see the rapid growth of AI and machine learning. The development of this technology has come a long way since its inception, and it continues to surprise us every day with its capabilities.

From creating stunning artwork with Midjourney and stable diffusion to providing human-like conversations with language models like ChatGPT, the potential applications of AI are seemingly limitless.

However, it's important to remember that while this technology has advanced significantly, there's still much to be learned and explored. As AI professionals, it's our job to guide companies through the exciting but sometimes daunting process of adopting this transformative technology, starting with datasets and ending with model deployment.

Even these fancy and huge new models require customization, and that’s where some of the work we do happen. However, a lot more of the work we do is completely custom.

We’ve been at this for several years, and I’m confident we can continue to help organizations unlock the full potential of AI as the technology matures, and continue pushing the boundaries of what's possible.

In which industries do you believe implementing artificial intelligence would have the most significant potential for success and why?

I believe that AI has the potential to transform a wide range of industries, including aerospace, defense, healthcare, finance, retail, and manufacturing.

In aerospace and defense, AI can be used to improve very custom capabilities like radar, drones, command and control systems, drone swarms, and even modeling and simulation for mission planning.

In Healthcare, I predict that patient outcomes should benefit from AI by enabling earlier and more accurate diagnoses, and new treatment options discovered or optimized using AI. There have also been some big failures in the field of AI and healthcare, and so this is an area where we see a lot of justified caution when adopting AI technology.

In finance, you hear a lot about AI for fraud detection and risk management, but the applications I actually see are more geared towards trading signal detection and portfolio management. I find that real-time trading is very hard to break into.

We see projects across many verticals, so I’ll just briefly mention that retail, education, and manufacturing are also interesting verticals.

What types of jobs are at risk of automation? What steps should they take to address potential job losses?

I like to think about the answer to this question in the context of Microsoft Excel in the 1990s. I’m sure excel killed a lot of jobs in pencil factories, pulp and paper mills, and in office buildings full of number crunchers who were replaced by spreadsheet gurus.

The lesson there is that every advance in technology deploys “creative destruction” into the economy. AI is a technology that improves efficiency, and so we see the creative destruction and think of all sorts of fears, but instead remember that we have a lot of work to do.

We need to cure cancer, increase longevity, and spread out into space. We humans can use all the help we can get.

With that in mind, it is a fair question. We should realize what jobs are at risk, and recognize that fewer people will work in those industries. There are several jobs that are at risk of automation, including repetitive tasks such as data entry, assembly line work, and customer service roles.

There will just be fewer jobs than before, and the work in those jobs will be more dynamic. Low-skilled manual labor, routine-based financial services, and administrative support functions will all still exist, but at a lower level of intensity because they will be competing with their more-automated counterparts.

I’m not a fan of recommendations like universal basic income as a solution to job losses. Instead, I think the universe is way too interesting and life is too short to retire at 30. I like the proposals to have governments support continuous learning, promote the adoption of transferable technical skills in the workforce, opening the economy through networking and open trade, and providing professional support for job seekers (like government-funded career counseling and job placement services).

It's important to note that while some jobs may be automated, new jobs will also be created as a result of the growth of the AI and automation industries. The key is to be proactive and adaptable, to ensure that workers remain valuable and employable in an ever-evolving job market.

Lemay.ai caters to various clients, from banking and finance to medical and pharmaceutical companies. What are the main challenges of adapting the framework and the model to each industry’s needs?

As an AI consulting company, the challenges we are called up to solve are those that our client’s couldn’t figure out on their own, and so we are naturally faced with extra-hard problems all the time.

Some of the main challenges we help with are data and data quality, systems design and requirements gathering, solution architecture for use-cases, integration with existing systems, privacy, security, and issues related to real-time systems like deployment on specialized hardware.

It’s a very diverse set of problems that we work on. No two days are the same. Addressing these challenges requires a combination of technical expertise, industry knowledge, and a deep understanding of the specific business requirements of each client.

At Lemay.ai, we work closely with our clients and their subject matter experts to understand their unique needs and to develop customized solutions that meet their specific requirements.

You also worked with the federal government and created AI-based solutions for the navy, aerospace and defense agencies. Can you tell us more about these projects and their significance?

Honestly, not really. We do try and publish press releases or case studies whenever possible, to push into the public domain some hints about the cool things we are working on. But mostly we can’t talk about the things we do for clients.

At Lemay.ai, the confidentiality of our clients is paramount. We take several steps to ensure the protection of client information. Specifically, we sign non-disclosure agreements, maintain on-premise secure storage, and access cloud-based resources using highly secure encryption, firewalls, and secure backups.

We have policies and procedures in place for access control to client data, and we get police checks and security clearance for employees so that they can work with sensitive data.

You have previously partnered with both Amazon and NVIDIA on various projects. From your experience, what are the main differences in approach and technologies when working with these two industry giants?

Unfortunately, there is not much I can say here either. IP is another critical component of the AI consulting business. Regarding who owns the things we make (intellectual property), we don’t claim ownership of the IP developed by our clients during the course of our work. As a general rule, our clients retain ownership of the IP that is developed during the course of our engagement.

Looking further into the future, how far are we from a true conscious artificial intelligence? Should we even strive to do it from an ethical standpoint?

We are very far from Artificial General Intelligence (AGI). Significant progress has been made in many areas of the field, and as a practitioner, it is exhausting to experience this rate of change in my everyday work.

A client asked us to write a literature review on natural language processing in 2023, and my response was “This will need to be updated within months” because this field is moving at warp speed. I have really enjoyed the “imagination” of emergent phenomena like deep-dream suddenly accelerated into full-on applications like stable diffusion.

It’s amazing to watch, but remember that this is more akin to a magic trick with playing cards than it is a smart general artificial intelligence. What you are seeing is a triumph of engineering like going from a horse to a car, but unlike the horse, the new AI “car” is not at all alive.

Rather than working on the development of consciousness, AI researchers are working on hiding the broken bits of the AI world from reaching the user.

From an ethical standpoint, I’m very unconvinced that conscious AI is possible within my lifetime. But never say never. We do need to continue to support universities and their scientists to investigate the issues, but private industry should not bother working on this.

It’s not that I think AI safety is not worthy of effort. It is. I’m specifically talking about AI ethics (see the book Superintelligence by Nick Bostrom). My argument is that those efforts should be publicly funded research, not something private industry focuses on.

There is real work to do that really is possible within my lifetime, and that’s where I believe we in the industry should focus our resources. We are likely to see AI regulations in the coming years, just as we saw many countries release their own AI roadmaps over the past few years.

Thank you for your time, Daniel Shapiro. Best of luck to you and Lemay.ai!

Keep up with Lemay.ai on LinkedIn, and check out some other successful businesses.