Tag Archive for: Chat GPT

Be careful! LinkedIn ‘rewrite with AI’ feature uses your expertise for free

If you’ve tried out LinkedIn’s ‘rewrite with AI’ feature – even like me, just to see what it looks like – you may be surprised to learn this.

LinkedIn 'rewrite with AI' button

In its FAQs, LinkedIn say they keep the information and “seek to minimise personal data” entered to train its generative AI software – this is in keeping with other services like Chat GPT.

LinkedIn 'rewrite with AI' T&Cs: The artificial intelligence models that LinkedIn uses to power generative AI features may be trained by LinkedIn or another provider. For example, some of our models are provided by Microsoft’s Azure OpenAI service. Where LinkedIn trains generative AI models, we seek to minimize personal data in the data sets used to train the models, including by using privacy enhancing technologies to redact or remove personal data from the training dataset. Your personal data may be used (or processed) for certain generative AI features on LinkedIn. For example, we process the data that you provide to the generative AI powered feature (e.g. writing suggestions), or that LinkedIn might create if you use a generative AI powered feature (e.g. profile writing suggestions). Like other features on LinkedIn, when you engage with generative AI powered features we process your interactions with the feature, which may include personal data (e.g., your inputs and resulting outputs, your usage information, your language preference, and any feedback you provide). See our Privacy Policy for more information about how we use and process personal data.

My former ITV News colleague, Suswati Basu points out that it also applies to the collaborative articles that you may see in your feed asking for your expert thoughts on a topic.

Example of a LinkedIn collaborative article

Some of you may be OK with that, but think about what you’re happy to give your expertise away for free – or include potentially sensitive information or names of people.

LinkedIn’s FAQs on its use of personal data and gen AI is here.

Read more: LinkedIn share algorithm tips on the best way to get engagement on posts

7 practical takeaways from Ofcom’s Online Nation report

How have our online habits changed this past year? Ofcom have done a thorough exploration of this, with their latest Online Nation report.

I’ve gone through this fascinating study to pull out what their findings mean for you – and how you can be inspired to better reach and engage your audiences.

1. Do you have a good YouTube strategy?

Ofcom Online Nation report survey: YouTube is Top social media services among UK adults

For too many, YouTube is still seen as a dumping ground for TV ads and other unloved videos. But it should really be a cornerstone of your digital comms strategy.

YouTube is now the most popular social or digital media platform among UK adults, overtaking Facebook.

Think about how you can provide helpful and engaging content on your most popular or searched for topics, with clear video titles that include relevant keywords, and interesting thumbnail images.

It’s no surprise that TikTok is popular among younger adults, with 18-24 year olds using it for an average of 55 minutes a day on there, but Snapchat is proving remarkably resilient too, with its users spending nearly an hour a day on there.

It feels too early to make any decisive views on Threads, but despite aggressive promotion on Meta’s other platforms, it’s not quite caught on so far. Like many of us, I’ve dipped in and out of it. But Channel 4 is doing a good job on there!

Don’t write off Twitter (or X) – although its usage has declined overall, it’s still the main place for breaking news, sport and other live events, if that’s what your audiences are interested in, and journalists and opinion formers are still active there.

 

2. The one size fits all social approach doesn’t work anymore

Overall, we’re seeing an increasing fragmentation of social media, with different groups of audiences using a wider range of channels. Just plonking the same post and content on Twitter, Facebook and Instagram is unlikely to work as it may have done before.

You need to examine what is working on what channel for you – even if it contradicts the general findings of this report – and even be bold about focusing on fewer of them.

Diversify your forms of engagement and communication too – we’ve seen how a whim of a social media giant or tweak in the Google algorithm can have a serious impact.

Also, think what you’e trying to achieve – is it to provide information and drive awareness, where social can be beneficial, or is it driving people to your website, which is becoming trickier through these platforms without paying?

Tailoring and optimising content and messaging is so important as well. Generally, we have less time and patience than ever before – if your caption or first few seconds don’t interest them, it’s likely to struggle.

I’ve proven this theory wrong, with some 16:9 videos on TikTok getting hundreds of thousand of views, but the content has to really resonate with, and engage, your followers.

And with the general shift to more algorithmic-based feeds, look into an organic promotion strategy to help boost the visibility of your posts – this doesn’t mean just influential voices, but your own staff too!

 

3. Is your content and site optimised for mobile?

This may sound silly in 2023, but we are using our phones more than ever to go online, with computer and tablet usage declining, according to Ofcom.

Even though those aged 55+ use smartphones the least, this still accounted for the vast majority of their time spent online.

The irony wasn’t lost as I tried to read Ofcom’s PDF-only report on my phone…

 

4. Consider reaching and engaging your audiences on WhatsApp

Apparently, we have an average of 36 apps on our phones – having culled a load recently from my phone that I hadn’t used in ages, this surprised me!

Meta-owned apps were the top ones used by adults, with WhatsApp still the favoured app of the nation.

Have you thought about creating a broadcast channel or explored whether you can gain access to Channels?

Success on there so far has been a bit hit and miss (my assessment is: less link posts, more native chat and content.)

 

5. Think about how AI can help your teams do their jobs better and provide a better service and experience for your audiences

I was fascinated to learn that as many as three in ten UK internet users had tried out a generative AI tool, such as Chat GPT, DALL-E or Snapchat’s AI. There’s an apparent willingness for people to use them, if they’re clearly labelled as such

But as Microsoft discovered, great care is needed when using it to avoid undermining customer confidence in what you do.

AI is also being used to provide great benefit to comms jobs – Buckingham Palace used AI-driven software to help assess the notoriously-tricky-to-measure online sentiment to the recent royal visit to Kenya, and Whitehall is testing their own in-house tool to see how it can help improve tasks such as in press offices.

I recommend following Peter Heneghan who’s doing exciting work in this space.

 

6. We all have a duty as content creators and publishers to protect our audiences

“Attitudes to being online are complex and often ambivalent”, notes Ofcom.

It adds that “adults recognise the benefits of being online at a personal level, but less so for society overall… Although being online provides both adults and children with many benefits, the online environment also has the potential to expose internet users to harm.”

Misinformation is the biggest issue – with those surveyed claiming political or electoral content is the biggest problem. A quarter of users had also encountered hateful, offensive or discriminatory content in the past month.

Think about the steps you can take to protect followers: do you have a clear guide in place? Do staff know about it, from a professional, and also sadly, personal basis?

Worryingly, with the shift to more algorithmic-based feeds, over a third of users said their most recent harmful interaction came for just scrolling through their feeds, or the for you page.

Combined with the fact that less than half of those surveyed were aware of platforms’ rules and safety measures in place, I’d love to see a high profile campaign by these apps, government and Ofcom to raise awareness of how to report potential issues, and also improve media literacy of users online.

 

7. We’re more online than ever – but remember those who aren’t

UK adults spend on average 3¾ hours online – slightly more than this time last year. It’s no surprise that younger people spend nearly an hour more, and 65+ an hour less.

But what struck me was the numbers that don’t have access to the internet at home, either through desire or affordability.

It’s 7% of 16+, rising to 18% of over-65s. Over a quarter of people said this was due to cost.

Despite the race to be more and more digital, don’t forget to include all of your audiences in your communication efforts.

You can read Ofcom’s fasinating Online Nation report here.

Microsoft’s AI error shows how we could lose customer confidence

Does a travel guide to Ottawa, seemingly written by AI, on Microsoft’s site really suggest visiting a local food bank – and doing so “on an empty stomach”?
Yes, yes it really does.
Microsoft AI travel guide article suggests going to Ottawa Food Bank "on an empty stomach"

Microsoft

The mistake, spotted by Paris Marx, highlights the concerns around the use of AI to produce content without having a knowledgable human editor overseeing its work.

It’s a shame as this misuse of an exciting technology can really set back confidence in it.

You can check the error for yourself here.

Interestingly, Ofcom research shows that three in ten UK internet users had tried out a generative AI tool, such as Chat GPT, DALL-E or Snapchat’s AI. There’s an apparent willingness for people to use them, if they’re clearly labelled as such.

This Microsoft AI error comes on the day that TUI have launched their own Chat GPT-style tour guide app, which also appears to have quite severe limitations, according to the Guardian:

“A test of the service highlighted its ‘experimental’ nature, as the bot struggled with a basic conversation.”

TUI AI chatbot
The Guardian

The app does make it clear its experimental and an AI bot – but if users experience faults with a service very quickly and easily, they are unlikely to want to use it again.

Ensuring sufficient testing of AI-driven projects before making them public is really important.

As a senior TUI executive points out, it can really help where real people use generative AI along with their own knowledge for tasks – such as in call centres.

We’re not at the stage where people’s jobs will be replaced by this technology – but if businesses harness this innovation, properly, it can make a real benefit to the customer experience.