AI safety

Hi!

It’s easy to get swept away with exciting new AI tools. The AI world is developing at such a fast pace that there are regularly shiny new features and possibilities to tempt us. 

The benefits can be huge but we also need to consider the potential risks and implications.

Beyond the basic check that an AI company is reputable, there should also be questions asked about privacy, copyright, ethics and more.

Here are some considerations to bear in mind when adopting new AI tools:

Data privacy and security

Before we feed these AI tools with all of our personal and business information to help with work problems or solve home dilemmas, we should be asking what happens to all of our data. 

How is your sensitive and personal data being used or shared and what processes are in place to ensure its safety? What regulations apply?

Training data bias

You get out what you put in: if there are biases present in the training data then the output will potentially reflect this. What are AI companies doing to combat and mitigate this? 

It is essential to be aware of the pitfalls when it comes to AI-generated output so that the output can be judged and used accordingly without simply perpetuating bias and discrimination.

Misinformation

You can’t take AI output as gospel. Errors can occur so it pays to be discerning and not just take information at face value. 

Any tool that is designed to be creative - such as a generative content model - may also be creative with the truth. ‘Facts’ may be based on pre-existing misinformation included in the training data or ‘quotes’ created to demonstrate a point. It’s crucial to check reputable sources and question the data provided.

Ethics

Earlier this year, we saw actor Scarlett Johansson (who had previously turned down an offer to work with OpenAI) take issue with their GPT-4o model, Sky. Her complaint? The AI's voice sounded uncannily like her own.

Is it ok for AI to imitate artists, writers and musicians with no credit or compensation to the original creators? At what point do AI-generated images, audio and other creative works infringe on the copyright of original creators if their work is being used without consent to train AI models?

We’re entering complex ethical and legal territory!

The role of humans

Certain tools and use cases are going to require more human intervention and oversight than others. For example, AI chatbots for health and fitness tools should ideally involve professional human oversight, with clear limitations on the topics the chatbot can advise on and discuss.

The question of responsibility and accountability for the output is important here. Does the company have human input on the information and advice given before it is presented to users? If not, what safeguards are in place to ensure users are not being given inaccurate or harmful information?

Overreliance

If AI tools come up with ideas for you, write for you, create for you, then you’re missing the opportunity to develop your cognitive skills, think critically and find your own, unique voice.

As the saying goes, if you don’t use it, you lose it. When we start depending on these tools we lose critical skills. Not exercising or neglecting basic strength and flexibility makes life physically harder as you age and it’s equally important to keep exercising your mind - studies have shown that mentally stimulating activities can help to lower the risk of cognitive decline.

AI tools should be used to enhance our abilities, rather than replace our own thinking and creativity.

If you want to learn more about AI safety then you may be interested in a LinkedIn Live event held next week by Heather Murray, founder of AI for Non-Techies. 

Heather is an AI expert and enthusiast who specializes in making AI accessible to everyone and on 30 July she will be chatting about the safe use of AI with industry experts.

Find out how AI tools use your data, what to look out for when considering new tools and how current data protection regulations apply to AI tools. 

On a more general note, if you’re keen to learn more about AI but feel your technical skills aren’t up to it or you’re put off by technical jargon then Heather is someone you should definitely start following.

OK, speak soon!

Inge
Top Apps Founder

P.S. Check out the Top Apps AI courses here.