Beyond AI: Blending intelligence for better outcomes
Artificial Intelligence (AI) has huge transformative potential for businesses, with 3 in 5 UK marketers viewing the technology as an opportunity for their organisation.[1]
Yet it is not well understood among consumers, with many only familiar with large language models (LLMs) like ChatGPT and ‘uncanny valley’ image generators like Midjourney.
This has led to negative perceptions that AI infringes human artists’ copyright, along with the broader opinion that deploying machine-learning AI in any context is unethical.
Despite this, AI is clearly here to stay. While the UK government’s consultation on copyright and AI[2] recently ended, businesses should not delay assessing the potential of AI as part of a broader business intelligence and marketing strategy.
More types of AI
LLMs are by no means the only AI models, although they are among the most widely known (and widely used) among consumers.
Some popular AI models include:
- Chatgpt
- Claude
- DeepSeek
- Gemini
More complex AI services include Perplexity, described as an “AI-powered Swiss Army Knife for information discovery and curiosity”[3], and Consensus, an AI search engine focused on academic research paper.
This is just a small selection of the diversity of platforms available in the AI ecosphere, and it can be quite fascinating to ask the same question to different models and compare their responses.
Consumer opinions of AI
Some consumers do have their doubts about the use of AI in marketing campaigns, but the tide is turning, and in many cases people just want there to be some human involvement.
For example, Statista reports that nearly three quarters (73%) of UK adults think fully automated AI marketing campaigns should be “carefully regulated”, with over two fifths (41%) saying they agree “strongly” with that statement.[4]
Indeed, one of the top objectives of AI deployment in UK marketing is to “serve customers better and improve loyalty” (18%), second only to improving internal efficiencies (21%).[7]
Uses for AI in marketing in the UK
As of the summer of 2024, the biggest application of AI in UK marketing was to enable more accurately targeted and personalised content.[8]
However, it’s crucial to keep the human touch in campaigns. As seen above, there is still a high rate of concern across all demographics when it comes to fully automated AI marketing.
This raises concerns for marketers: in a July 2023 survey spanning Europe and the Americas, 85% of marketing professionals said generative AI posed a risk for the safety of their brand’s reputation.[9]
With AI unlocking finer analysis of consumer data and routes to genuine per-person personalisation, brand marketers need to collaborate with a multi-resource marketing agency that understands this evolving technology and how to deploy it ethically.
Enhancing existing channels with AI
Some relatively primitive examples of AI have been used in marketing for some time. Website AI chatbots are one such case.
A chatbot can be trained with information about the relevant business, such as the different bases, toppings and sauces available on a pizza delivery menu. Customers can add their own data to this, including allergies and dietary preferences.
More sophisticated AI language models, equipped with voice recognition and text-to-speech capabilities, can turn this interaction into a spoken conversation, with the option to speak to a human if the chatbot cannot fully answer the query.
AI can automate boring and repetitive tasks to make existing channels more efficient in any number of ways, for example:
- Clarifying communications e.g. correcting emails
- Creating scripts to generate reports
- Explaining complex subjects like data encryption
- Researching new concepts introduced into the business
- Writing draft documents e.g. user instructions
When using AI in business communications, it’s important to use a platform designed to be secure. Consumer AI models might not offer the level of data protection necessary when inputting sensitive information about clients and stakeholders.
Pulling data into AI platforms
Most of the major AI platforms are growing over time, with the ability to import data from external sources – and this can remove some of the bumps and seams when connecting with data from individual client accounts.
SharePoint documents can be brought into Microsoft’s Copilot AI, allowing it to answer questions based on the contents of Word, Excel, PowerPoint and PDF files.
Google’s Gemini can also access uploaded documents and images, as well as (with the user’s express permission), other Google ‘apps’ like Calendar, Google Docs, Gmail and Search.
There is even a Google-owned AI model called NotebookLM capable of generating a podcast-style ‘audio overview’ of uploaded documents in a variety of formats.
As the AI bubble stabilises and solidifies at a sustainable size, the flood of poor-quality AI works we have seen so far will also find an equilibrium, with next-generation AI models capable of creating a higher standard of output.
We are already seeing this as new generations of AI ‘thought engines’ publish their logic and reasoning to show how they arrived at a conclusion.
Future potential of AI agents
Microsoft CEO Satya Nadella believes AI-powered ‘agents’ are the future of how we interact with technology – and will one day replace all conventional software.
Speaking on the BG2 podcast[10], he said: “I think the notion that business applications exist, that’s probably where they’ll all collapse in the agent era, because if you think about it, they are essentially CRUD databases with a bunch of business logic.”
AI enthusiast and YouTuber Matthew Berman[11] added: “Create, Read, Update, Destroy (CRUD), those are the base units of interactions between an interface and a database… so all he’s saying is there’s going to be the underlying database, and then it’s going to be agents that are interacting directly with the database.”
The implications of this are huge. AI agents interacting with the core data have the potential to eliminate the entire business application stack as we currently understand it.
Maximising efficiency and optimisation with AI
All of this comes back to optimising outcomes by blending AI with human ingenuity. It’s worth noting that much of the delay between submitting a query and receiving a response is artificial. The AI is quite capable of answering almost instantly, but most human users prefer the illusion of some ‘thinking time’.
When it’s running at full speed, AI can analyse vast quantities of data at a blistering pace, allowing complex reports to be integrated into marketing analytics dashboards in real-time.
This can underpin strategies like the ‘fail fast’ principle, where ideas are prototyped and tested and, if they’re found to be not working as hoped, they are ditched as quickly as possible.
AI is massively effective at this kind of strategy, as it can help you to brainstorm, refine and test out ideas, saving time and money while still having human input into the process.
Beyond the numbers
Promotional marketing is not just about generating sales – it’s also about building your brand, and AI still lacks the empathy to understand the way you connect with advocates and influencers.
By working with a multi-channel marketing expert like MRM, you can make sure your use of AI is more than just a numbers game, but also takes into account how people feel about your brand and products.
[1] https://www.statista.com/statistics/1459938/ai-opportunity-marketers-uk/
[2] https://www.gov.uk/government/consultations/copyright-and-artificial-intelligence
[3] https://www.perplexity.ai/hub/getting-started#what-is-perplexity
[4] https://www.statista.com/statistics/1396687/ai-marketing-campaigns-regulated-uk/
[5] https://www.statista.com/statistics/1396698/ai-marketing-campaigns-regulated-age-uk/
[6] https://www.statista.com/statistics/1396693/ai-marketing-campaigns-regulated-gender-uk/
[7] https://www.statista.com/statistics/1394319/objectives-ai-deplyment-marketing-uk/
[8] https://www.statista.com/statistics/1331064/applications-ai-digital-advertising-uk/
[9] https://www.statista.com/statistics/1405095/ai-risk-level-brand-safety/
Here are a few popular AI models:
In addition, there are models like Grok, used in X (ex Twitter), and LLaMA, used by Meta. There’s also the Perplexity, which uses several models https://www.perplexity.ai/
https://www.perplexity.ai/hub/getting-started#what-is-perplexity
Another interesting service is https://consensus.app/ which is based on academic papers and official research.
Statista sources:
https://www.statista.com/statistics/1459938/ai-opportunity-marketers-uk/
https://www.statista.com/statistics/1396687/ai-marketing-campaigns-regulated-uk/
https://www.statista.com/statistics/1405095/ai-risk-level-brand-safety/
https://www.statista.com/statistics/1394319/objectives-ai-deplyment-marketing-uk/
https://www.statista.com/statistics/1331064/applications-ai-digital-advertising-uk/
The Microsoft CEO predicts that AI agents will replace the software we use today. Here is a shortened statement by Satya Nadella with commentary: https://www.youtube.com/watch?v=uGOLYz2pgr8
…The application available https://www.deepseek.com/ is from China, but because it’s somewhat like an open-source model, Microsoft introduced (copied it) this model into its cloud.
Going beyond AI for optimal results:
https://momentumitsma.com/insights/rethinking-abm-outperforming-the-market-in-the-world-of-ai