Data Science and AI: Challenges and Opportunities

With AI's growing sophistication, many are worried that this life-changing technology will go too far. There are reports of people losing their jobs to AI, and some worry that robots will control us all if AI becomes too intelligent.

Particularly within the world of data science, there’s uneasiness given AI's ability to complete the daily tasks typically performed by data professionals. This risks pushing data scientists and other data experts in the "AI could replace me" camp. But things are more nuanced than that.

Why data scientists shouldn't be too concerned or "Exquisite Bullshitters"

Fact vs. fiction

Each one of us holds certain biases. Whether those biases lead us to right or wrong conclusions, we’re able to recognize our beliefs and adjust them based on new information. At the very least, we can continue to hold our opinions while admitting that they come with biases. Machines don't have this ability, and that’s causing some problems.

As author Ariel Aberg-Riger writing for the MIT Technology Review recently put it, machines are "Exquisite Bullshitters." They process information fed to them and regurgitate answers to our questions based on what they are told. They don't actually know what they're saying. What they churn out isn’t necessarily factual either; it's more that they're just getting better at making what they generate sound true.

Dan Vigdor of Forbes explained that “Generative AI tools like ChatGPT, Bard, and other forms of language modelling could present various threats to how information is generated, organized and made accessible by search engines like Google.” Those who control these tools can use them to advance harmful and misleading ideals by feeding them misinformation, and the machine itself can do nothing to stop it.

Most of the content fed into language-learning models finds its source in the open internet. This unfortunately ensures that the AI only has a limited pool of knowledge to draw from as only a fraction of the world’s knowledge is available online. Also, generated content may be prejudiced if the training dataset lacks balanced representation of data points.

Ethics isn't for the robots

In 2016, AI recreated a Rembrandt painting. It analyzed hundreds of the artists' works and seemingly created another masterpiece. But that begs the question: who gets the credit for this new work? Technically, Rembrandt himself didn’t make it. After all, he’s been dead for some 300-odd years, but the computer could have never recreated such a painting without the efforts of this gifted painter.

As UNESCO says, getting AI to generate creative work leads to questions surrounding ownership. With AI bringing in new definitions of what it means to be an author, we need human minds cultivated by eons of culturally sensitive thinking to find the best way forward - for both man and machine.

Ethical concerns shroud even facial recognition technology. One study found that African American and Asian faces were up to 100 times more likely to be misidentified than white faces and the highest false-positive rate was among Native Americans. Given that facial recognition disproportionately misidentifies black faces, this can lead to wrongful accusations and arrests of already overly surveilled groups.

Leave the "deep learning" to the humans

AI technology has improved systems across domains like healthcare, education, finance, real estate, and even cyber security. Data scientists and data analysts have to dig through enormous amounts of data to extract insights, and identify patterns and trends.

Some of these tasks can and are being automated through machines and machine learning technology: Assembling and merging data from numerous sources. Data cleaning and delivery to the appropriate locations. Automatically deploy prepared models. Finding specific trends in data. Producing versions of certain models.

Rather than stealing data jobs, these automations are simply evolving them. Automating the more basic tasks leaves more room for humans to do the job's critical thinking and more complex aspects. Combining AI with human problem-solving like this has the potential to empower, rather than threaten, data scientists' careers.

New possibilities

As technology rapidly evolves, it’s highly likely the work of data scientists will have to adapt. Data scientists will have to learn new tools to make themselves more flexible as the field of data science changes. AI will even give way to new roles such as:

Prompt Engineer

• Design and develop prompts or instructions for AI models.
• Collaborate with AI researchers and developers to create effective prompts.
• Utilize expertise in natural language processing and linguistics.
• Iterate and refine prompts to improve AI system performance.
• Analyze user feedback to enhance the relevance of AI-generated content.

AI Trainer

• Curate and annotate datasets for training AI models.
• Create input-output pairs to teach AI systems accurate responses.
• Fine-tune models using curated data to improve performance.
• Collaborate with machine learning engineers to implement training strategies.
• Ensure AI models produce contextually appropriate and reliable outputs.

AI Auditor

• Evaluate AI systems to ensure ethical and unbiased behaviour.
• Assess AI models for compliance with industry regulations and standards.
• Identify potential biases and fairness issues in AI-generated content.
• Develop and implement auditing methodologies to test AI performance.
• Collaborate with AI developers to address and rectify identified issues.

Those in data who start preparing for roles like these now will make themselves more employable when these types of careers become more prevalent in the market.

Trepidation remains

Despite all this, there's no question that when we see machines capable of replicating human work, we start to ask ourselves if we're the next victim. Known as "AI anxiety," the risk remains that certain jobs - especially creative and administrative ones - could be lost to AI. Goldman Sachs has reported that AI could replace the equivalent of 300 million jobs.

However, the idea that AI could could replace 300 million jobs doesn't mean it will or it should. AI isn’t a replacement for human creativity and ingenuity. The same rings true for data professionals. In automating more standard tasks, there’s more room for the human capacity for deep and critical thinking.

Moving forward together

Today, we have the opportunity to take advantage of AI and the possibilities it allows for people to go deeper into data. Data scientists and analysts don't have to pivot away from their daily tasks to exclusively perform data interpretation. However, leveraging the convenience of AI to evolve their positions can open doors to new possibilities.

Ready to take the leap into the fascinating world of data science? Explore our data programs and discover a new career path in just a few clicks.

At Lighthouse Labs, we exist to make tech-enabled change an opportunity for all. We believe in the power of artificial intelligence (AI) to help people and make our world a better place. We recognize that the responsible and ethical development and use of AI is paramount to its success in driving progress.