General Business Forum Brought to you by Salesforce

Is AI still sci-fi or could it help your small business?

  1. A processor chip.
    iStock/mkos83
    Melissa Tredinnick

    Melissa Tredinnick UKBF Regular Staff Member

    258 1
    3 |

    From medicine, finance and transport to fashion and fake news, artificial intelligence has the potential to affect almost every part of our lives. 

    It’s no longer a concept confined to science-fiction, but something with a real impact on people and businesses around the world. PwC predicts that by 2030, AI could contribute up to $15.7 trillion to the global economy, boosting GDP in Northern Europe by 9.9%. Other regions stand to benefit even more, with North America seeing a 14.5% boost, and GDP in China set to grow by 26.1%. 

    The practical uses of the technology are wider-ranging than they ever have been, and with that rapid progress has come increased public awareness – as well as growing anxiety that the technology is set to replace human jobs rather than aid them, and that it could be used for unethical purposes.

    OpenAI, a research organisation owned by Elon Musk, released the final version of its text-generating AI system called GPT-2 in November 2019 – despite concerns that it could be used to produce eerily convincing misinformation and spam, and even be “fine-tuned” to produce terrorist propaganda. 

    On the other hand, there’s clear potential for AI to have a positive impact on the world, helping us to solve important problems that we could not easily have worked out ourselves. 

    In February this year, for example, a group of researchers from the Massachusetts Institute of Technology reported that they had used AI to identify a new antibiotic that would be able to kill some of the world’s most dangerous bacteria. 

    For the most part, however, today’s big developments in AI often feel more relevant to large organisations than small businesses. To many people, the term can sound like little more than a futuristic buzzword.

    The complexity of the topic and the technical terminology around it could be partly to blame for that, with phrases like machine learning and deep learning often used interchangeably with AI, along with talk of neural networks, natural language processing, and so on.

    Paul Appshaw, co-founder of AI consultancy ALP-AI, says AI is “a broad term used to describe computational systems designed to loosely imitate human brain functionality”.

    Machine learning, he adds, is a specific branch of that field, where a computer learns to predict a result, having been shown various relevant examples.

    “Think of image recognition – let’s say for apples,” Paul says. “You would pass a machine-learning algorithm thousands of different pictures of apples, and it would learn from these what makes an apple an apple. 

    “Then, if you show it a new photo that contains an apple, it should be able to identify the apple in the image. This is quite different from classical coding, where all decision making is hard-coded by the programmer.”

    Deep learning is essentially a development of machine learning, which uses complex networks of information that are built up into layers, in a way that’s designed to mimic the processes of the human brain.

    It’s easy to see how machine and deep learning systems could have valuable real-life applications, allowing you to feed a computer relevant information, and have it analyse the data, predict results and inform your decisions much faster than a human could. 

    For businesses, arguably the biggest benefit of this technology is its ability to improve efficiency. 

    If you’re running a retail business, for example, you could use a machine learning system to pick up on patterns in your sales and predict customer demand – allowing you to effectively manage your stock.

    Other uses for AI could include analysing CRM data to help inform sales decisions and improve customer service, or providing recommendations on your website based on products users have already viewed.

    It can also be applied to accounting processes and financial forecasting, with many finance apps already using AI to categorise transactions and produce insights.

    While developments in language processing like GPT-2 might suggest a future where news articles, emails, essays and more can all be written by a robot, automated journalism is already being used, mainly to free up journalists’ time by handling routine reporting and manual tasks. 

    For example, the BBC uses an API called ‘Juicer’, which extracts content from around 850 RSS feeds and automatically tags and categorises the stories it pulls out. The New York Times has experimented with a text editing interface called ‘Editor’ which automatically adds annotations to articles as they’re written, allowing for faster research and fact-checking. 

    Other news outlets use text generators, mainly to write statistical and financial news stories that tend to be formulaic in nature. 

    It’s easy to see how this technology might be used to automate the writing of product descriptions for online retail or even the maintenance of your company’s website content with SEO in mind.

    But while AI is already a valuable tool for businesses, and is likely to become even more powerful in the future, there’s a risk that its current usefulness has been in some ways overplayed. 

    Computer systems can perform complex data analysis in a way that has potentially huge benefits for people and businesses, but they’re still far from thinking like humans, and there are limits to what they can do.

    Paul warns that as excitement about the benefits of AI grows, small businesses may need to watch out for companies trying to cash in on the hype by offering to “upgrade” their systems to machine-learning ones, promising “increased efficiencies and ultimately profits”.

    This may not be strictly necessary for many SMEs, and it could end up costing more than it’s really worth – especially as the tools themselves are usually open-access.

    “That’s not to say that they aren’t useful,” he adds. “But they are only relevant to a certain class of problem – namely those where you are trying to predict, or just learn behaviours. 

    “Certain small businesses could benefit a great deal from these kinds of services, but it completely depends what they’re doing.”

    If you want to implement AI in your business, make sure you know what you’re using it for. 

    Look at specific processes in your business that take up large amounts of time, or that need a lot of manual work. It may be possible for a computer to do those jobs for you.

    #0
  2. gpietersz

    gpietersz UKBF Ace Full Member

    1,454 348
    I have done some work in this area.

    As with previous technologies, calling it artificial intelligence is rather optimistic. Its a very long way human intelligence.

    The current excitement is around massive improvements to a limited (albeit important) classes of problems in pattern recognition (machine learning) and significant improvements to natural language processing.

    Its most useful if you have large data sets to train it on which means it is not often a good fit for small businesses. The technologies are being developed (and popularised) by companies lie Google who have massive datasets.

    At the same time the technologies from the previous waves of AI technologies are being overshadowed. They are still useful - I have things like expert systems in mind (my very first job was developing a prototype expert system so I have a bit of a soft spot for them).

    Genetic algorithms are closely related to more recent machine learning approaches, but, again are much less popular than they were.
     
    Posted: Mar 13, 2020 By: gpietersz Member since: Sep 10, 2019
    #2
    Ray Newman likes this.
  3. bodie007

    bodie007 UKBF Contributor Full Member

    32 5
    I am using chat bots on my websites and client web sites. They need to be the right sort of chat bot. I have seen great results. Increased email subscribers, increased appointments booked and in one case a 400% increase in low value front end sales from cold web site visitors.
     
    Posted: Mar 21, 2020 By: bodie007 Member since: Jul 4, 2008
    #3
    Ray Newman likes this.
  4. Nick Walsh Studios

    Nick Walsh Studios UKBF Regular Full Member

    153 19
    We use Diib AI for SEO and Digital Marketing it is awesome.
     
    Posted: Apr 12, 2020 By: Nick Walsh Studios Member since: Apr 12, 2020
    #4